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
15 use PublicInbox::MsgIter;
16 use PublicInbox::Address;
17 use PublicInbox::WwwStream;
18 use PublicInbox::Reply;
19 use PublicInbox::ViewDiff qw(flush_diff);
21 use POSIX qw(strftime);
22 use Time::Local qw(timegm);
23 use PublicInbox::Smsg qw(subject_normalized);
24 use constant COLS => 72;
25 use constant INDENT => ' ';
26 use constant TCHILD => '` ';
27 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
31 if ($eml) { # called by WwwStream::async_eml or getline
32 my $smsg = $ctx->{smsg};
33 $ctx->{smsg} = $ctx->{over}->next_by_mid(@{$ctx->{next_arg}});
34 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
35 "../${\mid_href($smsg->{mid})}/" : '';
36 my $hdr = $eml->header_obj;
37 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
38 multipart_text_as_html($eml, $ctx);
40 $$obuf .= '</pre><hr>';
41 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
43 } else { # called by WwwStream::async_next or getline
44 $ctx->{smsg}; # may be undef
48 # /$INBOX/$MESSAGE_ID/ for unindexed v1 inboxes
51 my $eml = delete $ctx->{eml} or return;
52 my $hdr = $eml->header_obj;
54 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
55 multipart_text_as_html($eml, $ctx);
57 $$obuf .= '</pre><hr>';
58 eval { $$obuf .= html_footer($ctx, $hdr) };
62 sub no_over_html ($) {
64 my $bref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return; # 404
65 $ctx->{eml} = PublicInbox::Eml->new($bref);
66 PublicInbox::WwwStream::response($ctx, 200, \&no_over_i);
69 # public functions: (unstable)
73 my $ibx = $ctx->{-inbox};
74 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
75 my $over = $ctx->{over} = $ibx->over or return no_over_html($ctx);
77 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
78 $ctx->{smsg} = $over->next_by_mid(@$next_arg) or return; # undef == 404
79 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
82 # /$INBOX/$MESSAGE_ID/#R
86 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
88 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
91 my $ibx = $ctx->{-inbox};
92 if (my $url = $ibx->{infourl}) {
93 $url = prurl($ctx->{env}, $url);
94 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
97 my ($arg, $link, $reply_to_all) =
98 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
99 if (ref($arg) eq 'SCALAR') {
100 return '<pre id=R>'.ascii_html($$arg).'</pre>';
103 # mailto: link only works if address obfuscation is disabled
107 * If your mail client supports setting the <b>In-Reply-To</b> header
108 via mailto: links, try the <a
109 href="$link">mailto: link</a>
113 push @$arg, '/path/to/YOUR_REPLY';
114 $arg = ascii_html(join(" \\\n ", '', @$arg));
117 id=R><b>Reply instructions:</b>
119 You may reply publicly to <a
120 href=#t>this message</a> via plain-text email
121 using any one of the following methods:
123 * Save the following mbox file, import it into your mail client,
124 and $reply_to_all from there: <a
127 Avoid top-posting and favor interleaved quoting:
129 href="$p_url">$p_url</a>
131 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
132 switches of git-send-email(1):
137 href="$se_url">$se_url</a>
144 my $refs = references($hdr);
148 sub fold_addresses ($) {
149 return $_[0] if length($_[0]) <= COLS;
150 # try to fold on commas after non-word chars before $lim chars,
151 # Try to get the "," preceded by ">" or ")", but avoid folding
152 # on the comma where somebody uses "Lastname, Firstname".
153 # We also try to keep the last and penultimate addresses in
154 # the list on the same line if possible, hence the extra \z
155 # Fall back to folding on spaces at $lim + 1 chars
156 my $lim = COLS - 8; # 8 = "\t" display width
157 my $too_long = $lim + 1;
158 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
160 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
163 .{$too_long,}?)(?:\s|\z)/xgo));
166 sub _hdr_names_html ($$) {
167 my ($hdr, $field) = @_;
168 my @vals = $hdr->header($field) or return '';
169 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
173 my ($nr, $singular, $plural) = @_;
174 return "0 $plural" if $nr == 0;
175 $nr == 1 ? "$nr $singular" : "$nr $plural";
178 # human-friendly format
179 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
181 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
182 # this is already inside a <pre>
184 my ($ctx, $smsg, $eml, $more) = @_;
185 my $subj = delete $smsg->{subject};
186 my $mid_raw = $smsg->{mid};
187 my $id = id_compress($mid_raw, 1);
189 my $root_anchor = $ctx->{root_anchor} || '';
191 my $obfs_ibx = $ctx->{-obfs_ibx};
193 $subj = '(no subject)' if $subj eq '';
194 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
195 $subj = '<b>'.ascii_html($subj).'</b>';
196 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
197 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
199 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
201 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
203 # Deleting these fields saves about 400K as we iterate across 1K msgs
204 delete @$smsg{qw(ts blob)};
206 my $hdr = $eml->header_obj;
207 my $from = _hdr_names_html($hdr, 'From');
208 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
209 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
210 my $upfx = $ctx->{-upfx};
211 my $mhref = $upfx . mid_href($mid_raw) . '/';
212 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
213 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
214 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
215 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
216 my ($tlen, $clen) = (length($to), length($cc));
218 if (($tlen + $clen) > COLS) {
219 $to_cc .= ' To: '.$to."\n" if $tlen;
220 $to_cc .= ' Cc: '.$cc."\n" if $clen;
223 $to_cc .= ' To: '.$to;
224 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
226 $to_cc .= ' Cc: '.$cc if $clen;
230 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
233 my $mapping = $ctx->{mapping};
234 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
235 my $href = $upfx . mid_href($irt) . '/';
236 my $html = ascii_html($irt);
237 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
241 # scan through all parts, looking for displayable text
242 $ctx->{mhref} = $mhref;
244 $eml->each_part(\&add_text_body, $ctx, 1);
248 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
249 "<a\nhref=\"$mhref\">permalink</a>" .
250 " <a\nhref=\"${mhref}raw\">raw</a>" .
251 " <a\nhref=\"${mhref}#R\">reply</a>";
254 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
255 $rv .= "\t[relevance $pct%]";
258 my $nested = 'nested';
263 $flat = "<b>$flat</b>";
265 $nested = "<b>$nested</b>";
267 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
268 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
269 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
274 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
278 sub pad_link ($$;$) {
279 my ($mid, $level, $s) = @_;
281 my $id = id_compress($mid, 1);
282 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
286 my ($mid_raw, $irt, $id, $ctx) = @_;
288 my $mapping = $ctx->{mapping} or return $rv;
290 my $mid_map = $mapping->{$mid_raw};
292 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
293 my ($attr, $node, $idx, $level) = @$mid_map;
294 my $children = $node->{children};
295 my $nr_c = scalar @$children;
298 # delete saves about 200KB on a 1K message thread
299 if (my $refs = delete $node->{references}) {
300 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
302 my $irt_map = $mapping->{$$irt} if defined $$irt;
303 if (defined $irt_map) {
304 $siblings = $irt_map->[1]->{children};
305 $nr_s = scalar(@$siblings) - 1;
306 $rv .= $pad . $irt_map->[0];
308 my $prev = $siblings->[$idx - 1];
309 my $pmid = $prev->{mid};
311 my $s = ($idx - 1). ' preceding siblings ...';
312 $rv .= pad_link($pmid, $level, $s);
313 } elsif ($idx == 2) {
314 my $ppmid = $siblings->[0]->{mid};
315 $rv .= $pad . $mapping->{$ppmid}->[0];
317 $rv .= $pad . $mapping->{$pmid}->[0];
320 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
321 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
322 $attr =~ s!\n\z!</b>\n!s;
323 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
324 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
327 my $cmid = $children->[0]->{mid};
328 $rv .= $pad . $mapping->{$cmid}->[0];
330 my $s = ($nr_c - 1). ' more replies';
331 $rv .= pad_link($cmid, $level + 1, $s);
332 } elsif (my $cn = $children->[1]) {
333 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
337 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
339 my $nmid = $next->{mid};
340 $rv .= $pad . $mapping->{$nmid}->[0];
341 my $nnext = $nr_s - $idx;
343 my $s = ($nnext - 1).' subsequent siblings';
344 $rv .= pad_link($nmid, $level, $s);
345 } elsif (my $nn = $siblings->[$idx + 2]) {
346 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
349 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
352 # non-recursive thread walker
353 sub walk_thread ($$$) {
354 my ($rootset, $ctx, $cb) = @_;
355 my @q = map { (0, $_, -1) } @$rootset;
357 my ($level, $node, $i) = splice(@q, 0, 3);
358 defined $node or next;
359 $cb->($ctx, $level, $node, $i) or return;
362 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
366 sub pre_thread { # walk_thread callback
367 my ($ctx, $level, $node, $idx) = @_;
368 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
369 skel_dump($ctx, $level, $node);
372 sub thread_eml_entry {
373 my ($ctx, $level, $smsg, $eml) = @_;
374 my ($beg, $end) = thread_adj_level($ctx, $level);
375 $beg . '<pre>' . eml_entry($ctx, $smsg, $eml, 0) . '</pre>' . $end;
378 sub next_in_queue ($$) {
379 my ($q, $ghost_ok) = @_;
381 my ($level, $smsg) = splice(@$q, 0, 2);
383 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
384 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
389 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
390 my ($ctx, $eml) = @_;
393 my ($level, $smsg) = delete @$ctx{qw(level smsg)};
394 if ($ctx->{nr} == 1) {
395 $ctx->{-title_html} = ascii_html($smsg->{subject});
396 $ctx->zmore($ctx->html_top);
398 return thread_eml_entry($ctx, $level, $smsg, $eml);
400 return unless exists($ctx->{skel});
401 my $ghost_ok = $ctx->{nr}++;
403 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
405 if (exists $smsg->{blob}) { # next message for cat-file
406 $ctx->{level} = $lvl;
409 # buffer the ghost entry and loop
410 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
412 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
413 $ctx->zmore(${delete($ctx->{skel})});
419 sub stream_thread ($$) {
420 my ($rootset, $ctx) = @_;
421 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
422 PublicInbox::WwwStream::aresponse($ctx, 200, \&stream_thread_i);
425 # /$INBOX/$MESSAGE_ID/t/
428 my $mid = $ctx->{mid};
429 my $ibx = $ctx->{-inbox};
430 my ($nr, $msgs) = $ibx->over->get_thread($mid);
431 return missing_thread($ctx) if $nr == 0;
432 my $skel = '<hr><pre>';
433 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
434 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
435 $skel .= "<b\nid=t>Thread overview:</b> ";
436 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
437 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
438 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
439 $skel .= "-- links below jump to the message on this page --\n";
440 $ctx->{-upfx} = '../../';
441 $ctx->{cur_level} = 0;
442 $ctx->{skel} = \$skel;
443 $ctx->{prev_attr} = '';
444 $ctx->{prev_level} = 0;
445 $ctx->{root_anchor} = anchor_for($mid);
446 $ctx->{mapping} = {};
447 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
450 my $rootset = thread_results($ctx, $msgs);
452 # reduce hash lookups in pre_thread->skel_dump
453 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
454 walk_thread($rootset, $ctx, \&pre_thread);
457 return stream_thread($rootset, $ctx) unless $ctx->{flat};
459 # flat display: lazy load the full message from smsg
460 $ctx->{msgs} = $msgs;
461 $ctx->{-html_tip} = '<pre>';
462 PublicInbox::WwwStream::aresponse($ctx, 200, \&thread_html_i);
465 sub thread_html_i { # PublicInbox::WwwStream::getline callback
466 my ($ctx, $eml) = @_;
468 my $smsg = $ctx->{smsg};
469 if (exists $ctx->{-html_tip}) {
470 $ctx->{-title_html} = ascii_html($smsg->{subject});
471 $ctx->zmore($ctx->html_top);
473 return eml_entry($ctx, $smsg, $eml, scalar @{$ctx->{msgs}});
475 while (my $smsg = shift @{$ctx->{msgs}}) {
476 return $smsg if exists($smsg->{blob});
478 my $skel = delete($ctx->{skel}) or return; # all done
484 sub multipart_text_as_html {
485 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
487 # scan through all parts, looking for displayable text
488 $_[0]->each_part(\&add_text_body, $_[1], 1);
491 sub submsg_hdr ($$) {
492 my ($ctx, $eml) = @_;
493 my $obfs_ibx = $ctx->{-obfs_ibx};
494 my $rv = $ctx->{obuf};
496 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
497 my @v = $eml->header($h);
499 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
506 sub attach_link ($$$$;$) {
507 my ($ctx, $ct, $p, $fn, $err) = @_;
508 my ($part, $depth, $idx) = @$p;
510 # Eml iteration clobbers multipart ->{bdy}, so do not offer
511 # downloads for 0-byte multipart attachments
512 return unless $part->{bdy};
514 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
515 my $size = bytes::length($part->body);
517 # hide attributes normally, unless we want to aid users in
518 # spotting MUA problems:
519 $ct =~ s/;.*// unless $err;
520 $ct = ascii_html($ct);
522 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
524 } elsif ($ct eq 'text/plain') {
529 my $rv = $ctx->{obuf};
530 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
533 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
536 $$rv .= "[-- Attachment #$idx: ";
537 my $ts = "Type: $ct, Size: $size bytes";
538 my $desc = $part->header('Content-Description') // $fn // '';
539 $desc = ascii_html($desc);
540 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
543 submsg_hdr($ctx, $part) if $part->{is_submsg};
548 sub add_text_body { # callback for each_part
550 my $upfx = $ctx->{mhref};
551 my $ibx = $ctx->{-inbox};
552 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
553 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
554 my ($part, $depth, $idx) = @$p;
555 my $ct = $part->content_type || 'text/plain';
556 my $fn = $part->filename;
557 my ($s, $err) = msg_part_text($part, $ct);
558 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
560 my $rv = $ctx->{obuf};
561 if ($part->{is_submsg}) {
562 submsg_hdr($ctx, $part);
566 # makes no difference to browsers, and don't screw up filename
567 # link generation in diffs with the extra '%0D'
570 # will be escaped to `•' in HTML
571 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
573 # always support diff-highlighting, but we can't linkify hunk
574 # headers for solver unless some coderepo are configured:
576 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
577 # diffstat anchors do not link across attachments or messages,
578 # -apfx is just a stable prefix for making diffstat anchors
579 # linkable to the first diff hunk w/o crossing attachments
580 $idx =~ tr!.!/!; # compatibility with previous versions
581 $ctx->{-apfx} = $upfx . $idx;
583 # do attr => filename mappings for diffstats in git diffs:
584 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
586 delete $ctx->{-long_path};
588 if ($ibx->{-repo_objs}) {
589 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
591 $spfx =~ s!/([^/]*)/\z!/!;
593 my $n_slash = $upfx =~ tr!/!/!;
596 } elsif ($n_slash == 1) {
598 } else { # nslash == 2
603 $ctx->{-spfx} = $spfx;
606 # some editors don't put trailing newlines at the end:
607 $s .= "\n" unless $s =~ /\n\z/s;
609 # split off quoted and unquoted blocks:
610 my @sections = PublicInbox::MsgIter::split_quotes($s);
611 undef $s; # free memory
612 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
613 # badly-encoded message with $err? tell the world about it!
614 attach_link($ctx, $ct, $p, $fn, $err);
617 foreach my $cur (@sections) {
619 # we use a <span> here to allow users to specify
620 # their own color for quoted text
621 $$rv .= qq(<span\nclass="q">);
622 $$rv .= $l->to_html($cur);
625 flush_diff($ctx, \$cur);
628 $$rv .= $l->to_html($cur);
630 undef $cur; # free memory
634 sub _msg_page_prepare_obuf {
635 my ($hdr, $ctx) = @_;
636 my $over = $ctx->{-inbox}->over;
637 my $obfs_ibx = $ctx->{-obfs_ibx};
639 my $mids = mids_for_index($hdr);
640 my $nr = $ctx->{nr}++;
641 if ($nr) { # unlikely
644 $ctx->{first_hdr} = $hdr;
647 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
649 $rv .= "<pre\nid=b>"; # anchor for body start
651 $ctx->{-upfx} = '../' if $over;
652 my @title; # (Subject[0], From[0])
653 for my $v ($hdr->header('From')) {
654 my @n = PublicInbox::Address::names($v);
656 $title[1] //= ascii_html(join(', ', @n));
658 obfuscate_addrs($obfs_ibx, $v);
659 obfuscate_addrs($obfs_ibx, $title[1]);
661 $rv .= "From: $v\n" if $v ne '';
663 foreach my $h (qw(To Cc)) {
664 for my $v ($hdr->header($h)) {
667 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
668 $rv .= "$h: $v\n" if $v ne '';
671 my @subj = $hdr->header('Subject');
673 my $v = ascii_html(shift @subj);
674 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
676 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
678 for $v (@subj) { # multi-Subject message :<
680 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
681 $rv .= "Subject: $v\n";
683 } else { # dummy anchor for thread skeleton at bottom of page
684 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
685 $title[0] = '(no subject)';
687 for my $v ($hdr->header('Date')) {
689 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
692 if (!$nr) { # first (and only) message, common case
693 $ctx->{-title_html} = join(' - ', @title);
694 $rv = $ctx->html_top . $rv;
696 if (scalar(@$mids) == 1) { # common case
697 my $mhtml = ascii_html($mids->[0]);
698 $rv .= "Message-ID: <$mhtml> ";
699 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
701 # X-Alt-Message-ID can happen if a message is injected from
702 # public-inbox-nntpd because of multiple Message-ID headers.
703 my $lnk = PublicInbox::Linkify->new;
705 for my $h (qw(Message-ID X-Alt-Message-ID)) {
706 $s .= "$h: $_\n" for ($hdr->header_raw($h));
708 $lnk->linkify_mids('..', \$s, 1);
711 $rv .= _parent_headers($hdr, $over);
717 qq(expand[<a\nhref="T/#u">flat</a>) .
718 qq(|<a\nhref="t/#u">nested</a>] ) .
719 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
720 qq(<a\nhref="t.atom">Atom feed</a>);
723 sub thread_skel ($$$) {
724 my ($skel, $ctx, $hdr) = @_;
725 my $mid = mids($hdr)->[0];
726 my $ibx = $ctx->{-inbox};
727 my ($nr, $msgs) = $ibx->over->get_thread($mid);
728 my $parent = in_reply_to($hdr);
729 $$skel .= "\n<b>Thread overview: </b>";
731 if (defined $parent) {
732 $$skel .= SKEL_EXPAND."\n ";
733 $$skel .= ghost_parent('../', $parent) . "\n";
735 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
737 $ctx->{next_msg} = undef;
738 $ctx->{parent_msg} = $parent;
743 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
745 # nb: mutt only shows the first Subject in the index pane
746 # when multiple Subject: headers are present, so we follow suit:
747 my $subj = $hdr->header('Subject') // '';
748 $subj = '(no subject)' if $subj eq '';
749 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
751 $ctx->{prev_attr} = '';
752 $ctx->{prev_level} = 0;
753 $ctx->{skel} = $skel;
755 # reduce hash lookups in skel_dump
756 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
757 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
759 $ctx->{parent_msg} = $parent;
762 sub _parent_headers {
763 my ($hdr, $over) = @_;
765 my @irt = $hdr->header_raw('In-Reply-To');
768 my $lnk = PublicInbox::Linkify->new;
769 $rv .= "In-Reply-To: $_\n" for @irt;
770 $lnk->linkify_mids('..', \$rv);
772 $refs = references($hdr);
773 my $irt = pop @$refs;
775 my $html = ascii_html($irt);
776 my $href = mid_href($irt);
777 $rv .= "In-Reply-To: <";
778 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
782 # do not display References: if search is present,
783 # we show the thread skeleton at the bottom, instead.
786 $refs //= references($hdr);
788 @$refs = map { linkify_ref_no_over($_) } @$refs;
789 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
794 # returns a string buffer via ->getline
796 my ($ctx, $hdr) = @_;
797 my $ibx = $ctx->{-inbox};
799 my $skel = " <a\nhref=\"$upfx\">index</a>";
803 thread_skel(\$skel, $ctx, $hdr);
808 if (my $n = $ctx->{next_msg}) {
810 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
813 my $par = $ctx->{parent_msg};
818 if (my $p = $ctx->{prev_msg}) {
819 $prev = mid_href($p);
820 if ($p && $par && $p eq $par) {
821 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
822 'rel=prev>prev parent</a>';
825 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
827 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
829 } elsif ($u) { # unlikely
830 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
832 $rv .= "$next $prev$parent ";
834 $rv .= qq(<a\nhref="#R">reply</a>);
837 $rv .= msg_reply($ctx, $hdr);
840 sub linkify_ref_no_over {
842 my $href = mid_href($mid);
843 my $html = ascii_html($mid);
844 "<<a\nhref=\"../$href/\">$html</a>>";
849 'm' . id_compress($msgid, 1);
853 my ($upfx, $mid) = @_;
855 my $href = mid_href($mid);
856 my $html = ascii_html($mid);
857 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
862 $level ? INDENT x ($level - 1) : '';
866 my ($ctx, $level, $node, $idx) = @_;
867 ++$ctx->{root_idx} if $level == 0;
868 if ($node->{mid} eq $ctx->{mid}) {
869 $ctx->{found_mid_at} = $ctx->{root_idx};
875 sub strict_loose_note ($) {
878 " -- strict thread matches above, loose matches on Subject: below --\n";
880 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
882 " -- use mbox.gz link to download all $nr messages --\n";
888 my ($ctx, $msgs) = @_;
889 require PublicInbox::SearchThread;
890 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
892 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
893 # and preserved in the future. This bug is hidden by `sid' matches
894 # in get_thread, so we never noticed it until now. And even when
895 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
896 # bump since reindexing is expensive and users may not do it
898 # loose threading could've returned too many results,
899 # put the root the message we care about at the top:
900 my $mid = $ctx->{mid};
901 if (defined($mid) && scalar(@$rootset) > 1) {
902 $ctx->{root_idx} = -1;
903 my $nr = scalar @$msgs;
904 walk_thread($rootset, $ctx, \&find_mid_root);
905 my $idx = $ctx->{found_mid_at};
906 if (defined($idx) && $idx != 0) {
907 my $tip = splice(@$rootset, $idx, 1);
908 @$rootset = reverse @$rootset;
909 unshift @$rootset, $tip;
910 $ctx->{sl_note} = strict_loose_note($nr);
918 require PublicInbox::ExtMsg;
919 PublicInbox::ExtMsg::ext_msg($ctx);
923 my ($prev_subj, $subj, $val) = @_;
925 my $omit; # '"' denotes identical text omitted
926 my (@prev_pop, @curr_pop);
927 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
928 push(@prev_pop, pop(@$prev_subj));
929 push(@curr_pop, pop(@$subj));
932 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
933 if (scalar(@curr_pop) == 1) {
935 push @$prev_subj, @prev_pop;
936 push @$subj, @curr_pop;
941 sub skel_dump { # walk_thread callback
942 my ($ctx, $level, $smsg) = @_;
943 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
945 my $skel = $ctx->{skel};
946 my $cur = $ctx->{cur};
947 my $mid = $smsg->{mid};
949 if ($level == 0 && $ctx->{skel_dump_roots}++) {
950 $$skel .= delete($ctx->{sl_note}) || '';
953 my $f = ascii_html($smsg->{from_name});
954 my $obfs_ibx = $ctx->{-obfs_ibx};
955 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
957 my $d = fmt_ts($smsg->{ds});
958 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
959 if (exists $ctx->{searchview}) {
960 if (defined(my $pct = $smsg->{pct})) {
961 $d .= (sprintf(' % 2u', $pct) . '%');
967 $d .= ' ' . indent_for($level) . th_pfx($level);
969 $ctx->{first_level} ||= $level;
971 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
972 $ctx->{prev_attr} = $attr;
974 $ctx->{prev_level} = $level;
979 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
980 "$attr [this message]</a></b>\n";
983 $ctx->{prev_msg} = $mid;
986 $ctx->{next_msg} ||= $mid;
989 # Subject is never undef, this mail was loaded from
990 # our Xapian which would've resulted in '' if it were
991 # really missing (and Filter rejects empty subjects)
992 my @subj = split(/ /, subject_normalized($smsg->{subject}));
993 # remove common suffixes from the subject if it matches the previous,
994 # so we do not show redundant text at the end.
995 my $prev_subj = $ctx->{prev_subj} || [];
996 $ctx->{prev_subj} = [ @subj ];
997 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1000 my $subj = join(' ', @subj);
1001 $subj = ascii_html($subj);
1002 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1003 $end = "$subj</a> $omit$f\n"
1009 my $mapping = $unmatched ? undef : $ctx->{mapping};
1011 my $map = $mapping->{$mid};
1012 $id = id_compress($mid, 1);
1014 $map->[0] = "$d<a\nhref=\"$m\">$end";
1017 $m = $ctx->{-upfx}.mid_href($mid).'/';
1019 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1024 my ($ctx, $level, $node) = @_;
1026 my $mid = $node->{mid};
1027 my $d = ' [not found] ';
1028 $d .= ' ' if exists $ctx->{searchview};
1029 $d .= indent_for($level) . th_pfx($level);
1030 my $upfx = $ctx->{-upfx};
1031 my $href = $upfx . mid_href($mid) . '/';
1032 my $html = ascii_html($mid);
1034 my $mapping = $ctx->{mapping};
1035 my $map = $mapping->{$mid} if $mapping;
1037 my $id = id_compress($mid, 1);
1038 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1039 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1041 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1043 ${$ctx->{skel}} .= $d;
1049 (eval { $a->topmost->{ds} } || 0) <=>
1050 (eval { $b->topmost->{ds} } || 0)
1054 # accumulate recent topics if search is supported
1055 # returns 200 if done, 404 if not
1056 sub acc_topic { # walk_thread callback
1057 my ($ctx, $level, $smsg) = @_;
1058 my $mid = $smsg->{mid};
1059 my $has_blob = $smsg->{blob} // do {
1060 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1061 %$smsg = (%$smsg, %$by_mid);
1066 my $subj = subject_normalized($smsg->{subject});
1067 $subj = '(no subject)' if $subj eq '';
1068 my $ds = $smsg->{ds};
1069 if ($level == 0) { # new, top-level topic
1070 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1071 $ctx->{-cur_topic} = $topic;
1072 push @{$ctx->{order}}, $topic;
1076 # continue existing topic
1077 my $topic = $ctx->{-cur_topic}; # should never be undef
1078 $topic->[0] = $ds if $ds > $topic->[0];
1079 $topic->[1]++; # bump N+ message counter
1080 my $seen = $topic->[2];
1081 if (scalar(@$topic) == 3) { # parent was a ghost
1082 push @$topic, $subj;
1083 } elsif (!defined($seen->{$subj})) {
1084 push @$topic, $level, $subj; # @extra messages
1086 $seen->{$subj} = $mid; # latest for subject
1087 } else { # ghost message
1088 return 1 if $level != 0; # ignore child ghosts
1089 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1090 push @{$ctx->{order}}, $topic;
1097 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1099 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1104 my $ibx = $ctx->{-inbox};
1105 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1107 # sort by recency, this allows new posts to "bump" old topics...
1108 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1109 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1111 next unless defined $top_subj; # ghost topic
1112 my $mid = delete $seen->{$top_subj};
1113 my $href = mid_href($mid);
1114 my $prev_subj = [ split(/ /, $top_subj) ];
1115 $top_subj = ascii_html($top_subj);
1118 # $n isn't the total number of posts on the topic,
1119 # just the number of posts in the current results window
1123 $anchor = '#u'; # top of only message
1125 $n = " ($n+ messages)";
1126 $anchor = '#t'; # thread skeleton
1129 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1130 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1131 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1132 " $ds UTC $n - $mbox / $atom\n";
1133 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1134 my $level = $extra[$i];
1135 my $subj = $extra[$i + 1]; # already normalized
1136 $mid = delete $seen->{$subj};
1137 my @subj = split(/ /, $subj);
1138 my @next_prev = @subj; # full copy
1139 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1140 $prev_subj = \@next_prev;
1141 $subj = join(' ', @subj);
1142 $subj = ascii_html($subj);
1143 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1144 $href = mid_href($mid);
1145 $s .= indent_for($level) . TCHILD;
1146 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1150 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1154 # only for the t= query parameter passed to overview DB
1155 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1158 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1159 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1162 sub pagination_footer ($$) {
1163 my ($ctx, $latest) = @_;
1164 delete $ctx->{qp} or return;
1165 my $next = $ctx->{next_page} || '';
1166 my $prev = $ctx->{prev_page} || '';
1168 $next = $next ? "$next " : ' ';
1169 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1171 "<hr><pre>page: $next$prev</pre>";
1174 sub index_nav { # callback for WwwStream::getline
1176 return $ctx->html_top if exists $ctx->{-html_tip};
1177 pagination_footer($ctx, '.')
1180 sub paginate_recent ($$) {
1181 my ($ctx, $lim) = @_;
1182 my $t = $ctx->{qp}->{t} || '';
1183 my $opts = { limit => $lim };
1184 my ($after, $before);
1186 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1187 # if only $after exists "YYYYMMDD.." because "." could be skipped
1188 # if interpreted as an end-of-sentence
1189 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1190 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1192 my $ibx = $ctx->{-inbox};
1193 my $msgs = $ibx->recent($opts, $after, $before);
1194 my $nr = scalar @$msgs;
1195 if ($nr < $lim && defined($after)) {
1196 $after = $before = undef;
1197 $msgs = $ibx->recent($opts);
1198 $nr = scalar @$msgs;
1200 my $more = $nr == $lim;
1201 my ($newest, $oldest);
1203 $newest = $msgs->[0]->{ts};
1204 $oldest = $msgs->[-1]->{ts};
1205 # if we only had $after, our SQL query in ->recent ordered
1206 if ($newest < $oldest) {
1207 ($oldest, $newest) = ($newest, $oldest);
1208 $more = 0 if defined($after) && $after < $oldest;
1211 if (defined($oldest) && $more) {
1212 my $s = ts2str($oldest);
1213 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1215 if (defined($newest) && (defined($before) || defined($after))) {
1216 my $s = ts2str($newest);
1217 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1224 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1226 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1228 PublicInbox::WwwStream::response($ctx, dump_topics($ctx), \&index_nav);
1231 sub thread_adj_level {
1232 my ($ctx, $level) = @_;
1234 my $max = $ctx->{cur_level};
1236 return ('', '') if $max == 0; # flat output
1238 # reset existing lists
1239 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1240 $ctx->{cur_level} = 0;
1242 } elsif ($level == $max) { # continue existing list
1244 } elsif ($level < $max) {
1245 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1246 $ctx->{cur_level} = $level;
1247 ("$beg<li>", '</li>');
1248 } else { # ($level > $max) # start a new level
1249 $ctx->{cur_level} = $level;
1250 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1255 sub ghost_index_entry {
1256 my ($ctx, $level, $node) = @_;
1257 my ($beg, $end) = thread_adj_level($ctx, $level);
1258 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})