1 # Copyright (C) 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 List::Util qw(max);
10 use Text::Wrap qw(wrap); # stdlib, we need Perl 5.6+ for $huge
11 use PublicInbox::MsgTime qw(msg_datestamp);
12 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl mid_href
14 use PublicInbox::Linkify;
15 use PublicInbox::MID qw(id_compress mids mids_for_index references
17 use PublicInbox::MsgIter;
18 use PublicInbox::Address;
19 use PublicInbox::WwwStream qw(html_oneshot);
20 use PublicInbox::Reply;
21 use PublicInbox::ViewDiff qw(flush_diff);
23 use POSIX qw(strftime);
24 use Time::Local qw(timegm);
25 use PublicInbox::Smsg qw(subject_normalized);
26 use PublicInbox::ContentHash qw(content_hash);
27 use constant COLS => 72;
28 use constant INDENT => ' ';
29 use constant TCHILD => '` ';
30 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
34 if ($eml) { # called by WwwStream::async_eml or getline
35 my $smsg = $ctx->{smsg};
36 my $over = $ctx->{ibx}->over;
37 $ctx->{smsg} = $over ? $over->next_by_mid(@{$ctx->{next_arg}})
39 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
40 "../${\mid_href($smsg->{mid})}/" : '';
41 my $obuf = _msg_page_prepare_obuf($eml, $ctx);
43 multipart_text_as_html($eml, $ctx);
44 $$obuf .= '</pre><hr>';
47 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
49 } else { # called by WwwStream::async_next or getline
50 $ctx->{smsg}; # may be undef
54 # /$INBOX/$MSGID/ for unindexed v1 inboxes
55 sub no_over_html ($) {
57 my $bref = $ctx->{ibx}->msg_by_mid($ctx->{mid}) or return; # 404
58 my $eml = PublicInbox::Eml->new($bref);
60 PublicInbox::WwwStream::init($ctx);
61 my $obuf = _msg_page_prepare_obuf($eml, $ctx);
63 multipart_text_as_html($eml, $ctx);
64 $$obuf .= '</pre><hr>';
67 eval { $$obuf .= html_footer($ctx, $eml) };
68 html_oneshot($ctx, 200, $$obuf);
71 # public functions: (unstable)
75 my $ibx = $ctx->{ibx};
76 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
77 my $over = $ibx->over or return no_over_html($ctx);
79 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
81 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$next_arg) or
82 return; # undef == 404
84 # allow user to easily browse the range around this message if
86 $ctx->{-t_max} = $smsg->{ts};
87 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
90 # /$INBOX/$MESSAGE_ID/#R
94 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
96 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
99 my $ibx = $ctx->{ibx};
100 if (my $url = $ibx->{infourl}) {
101 $url = prurl($ctx->{env}, $url);
102 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
105 my ($arg, $link, $reply_to_all) =
106 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
107 if (ref($arg) eq 'SCALAR') {
108 return '<pre id=R>'.ascii_html($$arg).'</pre>';
111 # mailto: link only works if address obfuscation is disabled
115 * If your mail client supports setting the <b>In-Reply-To</b> header
116 via mailto: links, try the <a
117 href="$link">mailto: link</a>
121 push @$arg, '/path/to/YOUR_REPLY';
122 $arg = ascii_html(join(" \\\n ", '', @$arg));
125 id=R><b>Reply instructions:</b>
127 You may reply publicly to <a
128 href=#t>this message</a> via plain-text email
129 using any one of the following methods:
131 * Save the following mbox file, import it into your mail client,
132 and $reply_to_all from there: <a
135 Avoid top-posting and favor interleaved quoting:
137 href="$p_url">$p_url</a>
139 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
140 switches of git-send-email(1):
145 href="$se_url">$se_url</a>
148 Be sure your reply has a <b>Subject:</b> header at the top and a blank line
149 before the message body.
155 my $refs = references($hdr);
159 sub fold_addresses ($) {
160 return $_[0] if length($_[0]) <= COLS;
161 # try to fold on commas after non-word chars before $lim chars,
162 # Try to get the "," preceded by ">" or ")", but avoid folding
163 # on the comma where somebody uses "Lastname, Firstname".
164 # We also try to keep the last and penultimate addresses in
165 # the list on the same line if possible, hence the extra \z
166 # Fall back to folding on spaces at $lim + 1 chars
167 my $lim = COLS - 8; # 8 = "\t" display width
168 my $too_long = $lim + 1;
169 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
171 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
174 .{$too_long,}?)(?:\s|\z)/xgo));
177 sub _hdr_names_html ($$) {
178 my ($hdr, $field) = @_;
179 my @vals = $hdr->header($field) or return '';
180 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
184 my ($nr, $singular, $plural) = @_;
185 return "0 $plural" if $nr == 0;
186 $nr == 1 ? "$nr $singular" : "$nr $plural";
189 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
190 # this is already inside a <pre>
192 my ($ctx, $eml) = @_;
193 my $smsg = delete $ctx->{smsg};
194 my $subj = delete $smsg->{subject};
195 my $mid_raw = $smsg->{mid};
196 my $id = id_compress($mid_raw, 1);
198 my $root_anchor = $ctx->{root_anchor} || '';
200 my $obfs_ibx = $ctx->{-obfs_ibx};
202 $subj = '(no subject)' if $subj eq '';
203 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
204 $subj = '<b>'.ascii_html($subj).'</b>';
205 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
206 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
208 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
210 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
212 # Deleting these fields saves about 400K as we iterate across 1K msgs
213 delete @$smsg{qw(ts blob)};
215 my $from = _hdr_names_html($eml, 'From');
216 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
217 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
218 my $upfx = $ctx->{-upfx};
219 my $mhref = $upfx . mid_href($mid_raw) . '/';
220 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
221 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
222 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
223 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
224 my ($tlen, $clen) = (length($to), length($cc));
226 if (($tlen + $clen) > COLS) {
227 $to_cc .= ' To: '.$to."\n" if $tlen;
228 $to_cc .= ' Cc: '.$cc."\n" if $clen;
231 $to_cc .= ' To: '.$to;
232 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
234 $to_cc .= ' Cc: '.$cc if $clen;
238 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
241 my $mapping = $ctx->{mapping};
242 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
243 my $href = $upfx . mid_href($irt) . '/';
244 my $html = ascii_html($irt);
245 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
249 # scan through all parts, looking for displayable text
250 $ctx->{mhref} = $mhref;
251 $ctx->{changed_href} = "#e$id"; # for diffstat "files? changed,"
253 $eml->each_part(\&add_text_body, $ctx, 1);
257 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
258 "<a\nhref=\"$mhref\">permalink</a>" .
259 " <a\nhref=\"${mhref}raw\">raw</a>" .
260 " <a\nhref=\"${mhref}#R\">reply</a>";
262 delete($ctx->{-qry}) and
263 $rv .= qq[ <a\nhref="${mhref}#related">related</a>];
266 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
267 $rv .= "\t[relevance $pct%]";
270 my $nested = 'nested';
274 $flat = "<b>$flat</b>";
276 $nested = "<b>$nested</b>";
278 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
279 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
280 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
285 # do we have more messages? start a new <pre> if so
286 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
290 sub pad_link ($$;$) {
291 my ($mid, $level, $s) = @_;
293 my $href = defined($mid) ?
294 ("<a\nhref=#r".id_compress($mid, 1).">($s)</a>\n") :
296 (' 'x19).indent_for($level).th_pfx($level).$href;
300 # my ($mapping, $mid) = @_;
301 ($_[0]->{$_[1] // \'bogus'} // [ "(?)\n" ])->[0];
305 my ($mid_raw, $irt, $id, $ctx) = @_;
307 my $mapping = $ctx->{mapping} or return $rv;
309 my $mid_map = $mapping->{$mid_raw};
311 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
312 my ($attr, $node, $idx, $level) = @$mid_map;
313 my $children = $node->{children};
314 my $nr_c = scalar @$children;
317 # delete saves about 200KB on a 1K message thread
318 if (my $refs = delete $node->{references}) {
319 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
321 my $irt_map = $mapping->{$$irt} if defined $$irt;
322 if (defined $irt_map) {
323 $siblings = $irt_map->[1]->{children};
324 $nr_s = scalar(@$siblings) - 1;
325 $rv .= $pad . $irt_map->[0];
327 my $prev = $siblings->[$idx - 1];
328 my $pmid = $prev->{mid};
330 my $s = ($idx - 1). ' preceding siblings ...';
331 $rv .= pad_link($pmid, $level, $s);
332 } elsif ($idx == 2) {
333 $rv .= $pad . _skel_hdr($mapping,
335 $siblings->[0]->{mid} : undef);
337 $rv .= $pad . _skel_hdr($mapping, $pmid);
340 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
341 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
342 $attr =~ s!\n\z!</b>\n!s;
343 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
344 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
347 my $cmid = $children->[0] ? $children->[0]->{mid} : undef;
348 $rv .= $pad . _skel_hdr($mapping, $cmid);
350 my $s = ($nr_c - 1). ' more replies';
351 $rv .= pad_link($cmid, $level + 1, $s);
352 } elsif (my $cn = $children->[1]) {
353 $rv .= $pad . _skel_hdr($mapping, $cn->{mid});
357 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
359 my $nmid = $next->{mid};
360 $rv .= $pad . _skel_hdr($mapping, $nmid);
361 my $nnext = $nr_s - $idx;
363 my $s = ($nnext - 1).' subsequent siblings';
364 $rv .= pad_link($nmid, $level, $s);
365 } elsif (my $nn = $siblings->[$idx + 2]) {
366 $rv .= $pad . _skel_hdr($mapping, $nn->{mid});
369 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
372 # non-recursive thread walker
373 sub walk_thread ($$$) {
374 my ($rootset, $ctx, $cb) = @_;
375 my @q = map { (0, $_, -1) } @$rootset;
377 my ($level, $node, $i) = splice(@q, 0, 3);
378 defined $node or next;
379 $cb->($ctx, $level, $node, $i) or return;
382 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
386 sub pre_thread { # walk_thread callback
387 my ($ctx, $level, $node, $idx) = @_;
388 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
389 skel_dump($ctx, $level, $node);
392 sub thread_eml_entry {
393 my ($ctx, $eml) = @_;
394 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
395 $beg . '<pre>' . eml_entry($ctx, $eml) . '</pre>' . $end;
398 sub next_in_queue ($$) {
399 my ($q, $ghost_ok) = @_;
401 my ($level, $smsg) = splice(@$q, 0, 2);
403 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
404 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
409 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
410 my ($ctx, $eml) = @_;
411 return thread_eml_entry($ctx, $eml) if $eml;
412 return unless exists($ctx->{skel});
413 my $ghost_ok = $ctx->{nr}++;
415 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
417 if (exists $smsg->{blob}) { # next message for cat-file
418 $ctx->{level} = $lvl;
419 if (!$ghost_ok) { # first non-ghost
420 $ctx->{-title_html} =
421 ascii_html($smsg->{subject});
422 $ctx->zmore($ctx->html_top);
426 # buffer the ghost entry and loop
427 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
429 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
430 $ctx->zmore(${delete($ctx->{skel})});
436 sub stream_thread ($$) {
437 my ($rootset, $ctx) = @_;
438 @{$ctx->{-queue}} = map { (0, $_) } @$rootset;
439 PublicInbox::WwwStream::aresponse($ctx, 200, \&stream_thread_i);
442 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
445 $ctx->{-upfx} = '../../';
446 my $mid = $ctx->{mid};
447 my $ibx = $ctx->{ibx};
448 my ($nr, $msgs) = $ibx->over->get_thread($mid);
449 return missing_thread($ctx) if $nr == 0;
451 # link $INBOX_DIR/description text to "index_topics" view around
452 # the newest message in this thread
453 my $t = ts2str($ctx->{-t_max} = max(map { delete $_->{ts} } @$msgs));
454 my $t_fmt = fmt_ts($ctx->{-t_max});
456 my $skel = '<hr><pre>';
457 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
460 href="../../?t=$t">~$t_fmt UTC</a> | <a
461 href="../../">newest</a>]
464 $skel .= "<b\nid=t>Thread overview:</b> ";
465 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
466 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
467 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
468 $skel .= "-- links below jump to the message on this page --\n";
469 $ctx->{cur_level} = 0;
470 $ctx->{skel} = \$skel;
471 $ctx->{prev_attr} = '';
472 $ctx->{prev_level} = 0;
473 $ctx->{root_anchor} = 'm' . id_compress($mid, 1);
474 $ctx->{mapping} = {}; # mid -> [ header_summary, node, idx, level ]
475 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
478 my $rootset = thread_results($ctx, $msgs);
480 # reduce hash lookups in pre_thread->skel_dump
481 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
482 walk_thread($rootset, $ctx, \&pre_thread);
485 return stream_thread($rootset, $ctx) unless $ctx->{flat};
487 # flat display: lazy load the full message from smsg
488 $ctx->{msgs} = $msgs;
489 $ctx->{-html_tip} = '<pre>';
490 PublicInbox::WwwStream::aresponse($ctx, 200, \&thread_html_i);
493 sub thread_html_i { # PublicInbox::WwwStream::getline callback
494 my ($ctx, $eml) = @_;
496 my $smsg = $ctx->{smsg};
497 if (exists $ctx->{-html_tip}) {
498 $ctx->{-title_html} = ascii_html($smsg->{subject});
499 $ctx->zmore($ctx->html_top);
501 return eml_entry($ctx, $eml);
503 while (my $smsg = shift @{$ctx->{msgs}}) {
504 return $smsg if exists($smsg->{blob});
506 my $skel = delete($ctx->{skel}) or return; # all done
512 sub multipart_text_as_html {
513 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
515 # scan through all parts, looking for displayable text
516 $_[0]->each_part(\&add_text_body, $_[1], 1);
519 sub submsg_hdr ($$) {
520 my ($ctx, $eml) = @_;
521 my $obfs_ibx = $ctx->{-obfs_ibx};
522 my $rv = $ctx->{obuf};
524 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
525 my @v = $eml->header($h);
527 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
534 sub attach_link ($$$$;$) {
535 my ($ctx, $ct, $p, $fn, $err) = @_;
536 my ($part, $depth, $idx) = @$p;
538 # Eml iteration clobbers multipart ->{bdy}, so do not offer
539 # downloads for 0-byte multipart attachments
540 return unless $part->{bdy};
542 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
543 my $size = length($part->body);
544 delete $part->{bdy}; # save memory
546 # hide attributes normally, unless we want to aid users in
547 # spotting MUA problems:
548 $ct =~ s/;.*// unless $err;
549 $ct = ascii_html($ct);
551 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
553 } elsif ($ct eq 'text/plain') {
558 my $rv = $ctx->{obuf};
559 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
562 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
565 $$rv .= "[-- Attachment #$idx: ";
566 my $ts = "Type: $ct, Size: $size bytes";
567 my $desc = $part->header('Content-Description') // $fn // '';
568 $desc = ascii_html($desc);
569 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
572 submsg_hdr($ctx, $part) if $part->{is_submsg};
577 sub add_text_body { # callback for each_part
579 my $upfx = $ctx->{mhref};
580 my $ibx = $ctx->{ibx};
581 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
582 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
583 my ($part, $depth, $idx) = @$p;
584 my $ct = $part->content_type || 'text/plain';
585 my $fn = $part->filename;
586 my ($s, $err) = msg_part_text($part, $ct);
587 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
589 my $rv = $ctx->{obuf};
590 if ($part->{is_submsg}) {
591 submsg_hdr($ctx, $part);
595 # makes no difference to browsers, and don't screw up filename
596 # link generation in diffs with the extra '%0D'
599 # will be escaped to `•' in HTML
600 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
602 # always support diff-highlighting, but we can't linkify hunk
603 # headers for solver unless some coderepo are configured:
605 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
606 # diffstat anchors do not link across attachments or messages,
607 # -apfx is just a stable prefix for making diffstat anchors
608 # linkable to the first diff hunk w/o crossing attachments
609 $idx =~ tr!.!/!; # compatibility with previous versions
610 $ctx->{-apfx} = $upfx . $idx;
612 # do attr => filename mappings for diffstats in git diffs:
613 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
615 delete $ctx->{-long_path};
617 # absolute URL (Atom feeds)
618 if ($ibx->{coderepo}) {
619 if (index($upfx, '//') >= 0) {
621 $spfx =~ s!/([^/]*)/\z!/!;
623 my $n_slash = $upfx =~ tr!/!/!;
626 } elsif ($n_slash == 1) {
628 } else { # nslash == 2
633 $ctx->{-spfx} = $spfx;
636 # split off quoted and unquoted blocks:
637 my @sections = PublicInbox::MsgIter::split_quotes($s);
638 undef $s; # free memory
639 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
640 # badly-encoded message with $err? tell the world about it!
641 attach_link($ctx, $ct, $p, $fn, $err);
644 delete $part->{bdy}; # save memory
645 foreach my $cur (@sections) {
647 # we use a <span> here to allow users to specify
648 # their own color for quoted text
649 $$rv .= qq(<span\nclass="q">);
650 $$rv .= $l->to_html($cur);
653 flush_diff($ctx, \$cur);
656 $$rv .= $l->to_html($cur);
658 undef $cur; # free memory
662 sub _msg_page_prepare_obuf {
663 my ($eml, $ctx) = @_;
664 my $have_over = !!$ctx->{ibx}->over;
665 my $obfs_ibx = $ctx->{-obfs_ibx};
666 $ctx->{obuf} = \(my $rv = '');
667 my $mids = mids_for_index($eml);
668 my $nr = $ctx->{nr}++;
669 if ($nr) { # unlikely
670 if ($ctx->{chash} eq content_hash($eml)) {
671 warn "W: BUG? @$mids not deduplicated properly\n";
675 "<pre>WARNING: multiple messages have this Message-ID\n</pre><pre>";
677 $ctx->{first_hdr} = $eml->header_obj;
678 $ctx->{chash} = content_hash($eml) if $ctx->{smsg}; # reused MID
679 $rv .= "<pre\nid=b>"; # anchor for body start
681 $ctx->{-upfx} = '../' if $have_over;
682 my @title; # (Subject[0], From[0])
683 for my $v ($eml->header('From')) {
684 my @n = PublicInbox::Address::names($v);
686 $title[1] //= ascii_html(join(', ', @n));
688 obfuscate_addrs($obfs_ibx, $v);
689 obfuscate_addrs($obfs_ibx, $title[1]);
691 $rv .= "From: $v\n" if $v ne '';
693 foreach my $h (qw(To Cc)) {
694 for my $v ($eml->header($h)) {
697 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
698 $rv .= "$h: $v\n" if $v ne '';
701 my @subj = $eml->header('Subject');
703 my $v = ascii_html(shift @subj);
704 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
706 $rv .= $have_over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
708 for $v (@subj) { # multi-Subject message :<
710 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
711 $rv .= "Subject: $v\n";
713 } else { # dummy anchor for thread skeleton at bottom of page
714 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $have_over;
715 $title[0] = '(no subject)';
717 for my $v ($eml->header('Date')) {
719 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
720 $rv .= qq{Date: $v\t<a\nhref="#r">[thread overview]</a>\n};
722 if (!$nr) { # first (and only) message, common case
723 $ctx->{-title_html} = join(' - ', @title);
724 $rv = $ctx->html_top . $rv;
727 $ctx->{-linkify} //= PublicInbox::Linkify->new;
728 if (scalar(@$mids) == 1) { # common case
729 my $mhtml = ascii_html($mids->[0]);
730 $rv .= qq[Message-ID: <$mhtml> (<a href="raw">raw</a>)\n];
732 # X-Alt-Message-ID can happen if a message is injected from
733 # public-inbox-nntpd because of multiple Message-ID headers.
735 for my $h (qw(Message-ID X-Alt-Message-ID)) {
736 $s .= "$h: $_\n" for ($eml->header_raw($h));
738 $ctx->{-linkify}->linkify_mids('..', \$s, 1);
741 _parent_headers($ctx, $eml);
747 qq(expand[<a\nhref="T/#u">flat</a>) .
748 qq(|<a\nhref="t/#u">nested</a>] ) .
749 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
750 qq(<a\nhref="t.atom">Atom feed</a>);
753 sub thread_skel ($$$) {
754 my ($skel, $ctx, $hdr) = @_;
755 my $mid = mids($hdr)->[0];
756 my $ibx = $ctx->{ibx};
757 my ($nr, $msgs) = $ibx->over->get_thread($mid);
758 my $parent = in_reply_to($hdr);
759 $$skel .= "\n<b>Thread overview: </b>";
761 if (defined $parent) {
762 $$skel .= SKEL_EXPAND."\n ";
763 $$skel .= ghost_parent('../', $parent) . "\n";
765 $$skel .= "<a\nid=r>[no followups]</a> ".
768 $ctx->{next_msg} = undef;
769 $ctx->{parent_msg} = $parent;
774 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
776 # nb: mutt only shows the first Subject in the index pane
777 # when multiple Subject: headers are present, so we follow suit:
778 my $subj = $hdr->header('Subject') // '';
779 $subj = '(no subject)' if $subj eq '';
780 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
782 $ctx->{prev_attr} = '';
783 $ctx->{prev_level} = 0;
784 $ctx->{skel} = $skel;
786 # reduce hash lookups in skel_dump
787 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
788 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
790 $ctx->{parent_msg} = $parent;
793 sub _parent_headers {
794 my ($ctx, $hdr) = @_;
795 my @irt = $hdr->header_raw('In-Reply-To');
799 $s .= "In-Reply-To: $_\n" for @irt;
800 $ctx->{-linkify}->linkify_mids('..', \$s);
801 ${$ctx->{obuf}} .= $s;
803 $refs = references($hdr);
804 my $irt = pop @$refs;
806 my $html = ascii_html($irt);
807 my $href = mid_href($irt);
808 ${$ctx->{obuf}} .= <<EOM;
809 In-Reply-To: <<a\nhref="../$href/">$html</a>>
814 # do not display References: if search is present,
815 # we show the thread skeleton at the bottom, instead.
816 return if $ctx->{ibx}->over;
818 $refs //= references($hdr);
820 $_ = linkify_ref_no_over($_) for @$refs;
821 ${$ctx->{obuf}} .= 'References: '. join("\n\t", @$refs) . "\n";
825 # returns a string buffer
827 my ($ctx, $hdr) = @_;
832 my $qry = delete $ctx->{-qry};
833 if ($qry && $ctx->{ibx}->isrch) {
834 my $q = ''; # search for either ancestor or descendent patches
835 for (@{$qry->{dfpre}}, @{$qry->{dfpost}}) {
836 chop if length > 7; # include 1 abbrev "older" patches
839 chop $q; # omit trailing SP
840 local $Text::Wrap::columns = COLS;
841 local $Text::Wrap::huge = 'overflow';
842 $q = wrap('', '', $q);
843 my $rows = ($q =~ tr/\n/\n/) + 1;
848 ><pre>find likely ancestor, descendant, or conflicting patches for <a
849 href=#t>this message</a>:
850 <textarea name=q cols=${\COLS} rows=$rows>$q</textarea>
851 <input type=submit value=search
852 />\t(<a href=${upfx}_/text/help/#search>help</a>)</pre></form>
855 if ($ctx->{ibx}->over) {
856 my $t = ts2str($ctx->{-t_max});
857 my $t_fmt = fmt_ts($ctx->{-t_max});
858 my $fallback = $related ? "\t" : "<a id=related>\t</a>";
860 ${fallback}other threads:[<a
861 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
862 href="$upfx">newest</a>]
865 thread_skel(\$skel, $ctx, $hdr);
870 if (my $n = $ctx->{next_msg}) {
872 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
875 my $par = $ctx->{parent_msg};
880 if (my $p = $ctx->{prev_msg}) {
881 $prev = mid_href($p);
882 if ($p && $par && $p eq $par) {
883 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
884 'rel=prev>prev parent</a>';
887 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
889 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
891 } elsif ($u) { # unlikely
892 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
894 $rv .= "$next $prev$parent ";
895 } else { # unindexed inboxes w/o over
896 $skel = qq( <a\nhref="$upfx">latest</a>);
898 $rv .= qq(<a\nhref="#R">reply</a>);
901 $rv .= $related // '';
902 $rv .= msg_reply($ctx, $hdr);
905 sub linkify_ref_no_over {
907 my $href = mid_href($mid);
908 my $html = ascii_html($mid);
909 "<<a\nhref=\"../$href/\">$html</a>>";
913 my ($upfx, $mid) = @_;
915 my $href = mid_href($mid);
916 my $html = ascii_html($mid);
917 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
922 $level ? INDENT x ($level - 1) : '';
926 my ($ctx, $level, $node, $idx) = @_;
927 ++$ctx->{root_idx} if $level == 0;
928 if ($node->{mid} eq $ctx->{mid}) {
929 $ctx->{found_mid_at} = $ctx->{root_idx};
930 return 0; # stop iterating
935 sub strict_loose_note ($) {
938 " -- strict thread matches above, loose matches on Subject: below --\n";
940 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
942 " -- use mbox.gz link to download all $nr messages --\n";
948 my ($ctx, $msgs) = @_;
949 require PublicInbox::SearchThread;
950 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
952 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
953 # and preserved in the future. This bug is hidden by `sid' matches
954 # in get_thread, so we never noticed it until now. And even when
955 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
956 # bump since reindexing is expensive and users may not do it
958 # loose threading could've returned too many results,
959 # put the root the message we care about at the top:
960 my $mid = $ctx->{mid};
961 if (defined($mid) && scalar(@$rootset) > 1) {
962 $ctx->{root_idx} = -1;
963 my $nr = scalar @$msgs;
964 walk_thread($rootset, $ctx, \&find_mid_root);
965 my $idx = $ctx->{found_mid_at};
966 if (defined($idx) && $idx != 0) {
967 my $tip = splice(@$rootset, $idx, 1);
968 @$rootset = reverse @$rootset;
969 unshift @$rootset, $tip;
970 $ctx->{sl_note} = strict_loose_note($nr);
978 require PublicInbox::ExtMsg;
979 PublicInbox::ExtMsg::ext_msg($ctx);
983 my ($prev_subj, $subj, $val) = @_;
985 my $omit; # '"' denotes identical text omitted
986 my (@prev_pop, @curr_pop);
987 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
988 push(@prev_pop, pop(@$prev_subj));
989 push(@curr_pop, pop(@$subj));
992 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
993 if (scalar(@curr_pop) == 1) {
995 push @$prev_subj, @prev_pop;
996 push @$subj, @curr_pop;
1001 sub skel_dump { # walk_thread callback
1002 my ($ctx, $level, $smsg) = @_;
1003 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
1005 my $skel = $ctx->{skel};
1006 my $cur = $ctx->{cur};
1007 my $mid = $smsg->{mid};
1009 if ($level == 0 && $ctx->{skel_dump_roots}++) {
1010 $$skel .= delete($ctx->{sl_note}) || '';
1013 my $f = ascii_html(delete $smsg->{from_name});
1014 my $obfs_ibx = $ctx->{-obfs_ibx};
1015 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
1017 my $d = fmt_ts($smsg->{ds});
1018 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
1019 if (exists $ctx->{searchview}) {
1020 if (defined(my $pct = $smsg->{pct})) {
1021 $d .= (sprintf(' % 2u', $pct) . '%');
1027 $d .= ' ' . indent_for($level) . th_pfx($level);
1029 $ctx->{first_level} ||= $level;
1031 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
1032 $ctx->{prev_attr} = $attr;
1034 $ctx->{prev_level} = $level;
1039 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
1040 "$attr [this message]</a></b>\n";
1043 $ctx->{prev_msg} = $mid;
1046 $ctx->{next_msg} ||= $mid;
1049 # Subject is never undef, this mail was loaded from
1050 # our Xapian which would've resulted in '' if it were
1051 # really missing (and Filter rejects empty subjects)
1052 my @subj = split(/ /, subject_normalized($smsg->{subject}));
1053 # remove common suffixes from the subject if it matches the previous,
1054 # so we do not show redundant text at the end.
1055 my $prev_subj = $ctx->{prev_subj} || [];
1056 $ctx->{prev_subj} = [ @subj ];
1057 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1060 my $subj = join(' ', @subj);
1061 $subj = ascii_html($subj);
1062 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1063 $end = "$subj</a> $omit$f\n"
1069 my $mapping = $unmatched ? undef : $ctx->{mapping};
1071 my $map = $mapping->{$mid};
1072 $id = id_compress($mid, 1);
1074 $map->[0] = "$d<a\nhref=\"$m\">$end";
1077 $m = $ctx->{-upfx}.mid_href($mid).'/';
1079 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1084 my ($ctx, $level, $node) = @_;
1086 my $mid = $node->{mid};
1087 my $d = ' [not found] ';
1088 $d .= ' ' if exists $ctx->{searchview};
1089 $d .= indent_for($level) . th_pfx($level);
1090 my $upfx = $ctx->{-upfx};
1091 my $href = $upfx . mid_href($mid) . '/';
1092 my $html = ascii_html($mid);
1094 my $mapping = $ctx->{mapping};
1095 my $map = $mapping->{$mid} if $mapping;
1097 my $id = id_compress($mid, 1);
1098 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1099 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1101 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1103 ${$ctx->{skel}} .= $d;
1109 (eval { $a->topmost->{ds} } || 0) <=>
1110 (eval { $b->topmost->{ds} } || 0)
1114 # accumulate recent topics if search is supported
1115 # returns 200 if done, 404 if not
1116 sub acc_topic { # walk_thread callback
1117 my ($ctx, $level, $smsg) = @_;
1118 my $mid = $smsg->{mid};
1119 my $has_blob = $smsg->{blob} // do {
1120 if (my $by_mid = $ctx->{ibx}->smsg_by_mid($mid)) {
1121 %$smsg = (%$smsg, %$by_mid);
1126 my $subj = subject_normalized($smsg->{subject});
1127 $subj = '(no subject)' if $subj eq '';
1128 my $ds = $smsg->{ds};
1129 if ($level == 0) { # new, top-level topic
1130 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1131 $ctx->{-cur_topic} = $topic;
1132 push @{$ctx->{order}}, $topic;
1136 # continue existing topic
1137 my $topic = $ctx->{-cur_topic}; # should never be undef
1138 $topic->[0] = $ds if $ds > $topic->[0];
1139 $topic->[1]++; # bump N+ message counter
1140 my $seen = $topic->[2];
1141 if (scalar(@$topic) == 3) { # parent was a ghost
1142 push @$topic, $subj;
1143 } elsif (!defined($seen->{$subj})) {
1144 push @$topic, $level, $subj; # @extra messages
1146 $seen->{$subj} = $mid; # latest for subject
1147 } else { # ghost message
1148 return 1 if $level != 0; # ignore child ghosts
1149 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1150 push @{$ctx->{order}}, $topic;
1157 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1159 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1164 my $obfs_ibx = $ctx->{ibx}->{obfuscate} ? $ctx->{ibx} : undef;
1165 if (my $note = delete $ctx->{t_note}) {
1166 push @out, $note; # "messages from ... to ..."
1168 # sort by recency, this allows new posts to "bump" old topics...
1169 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1170 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1172 next unless defined $top_subj; # ghost topic
1173 my $mid = delete $seen->{$top_subj};
1174 my $href = mid_href($mid);
1175 my $prev_subj = [ split(/ /, $top_subj) ];
1176 $top_subj = ascii_html($top_subj);
1179 # $n isn't the total number of posts on the topic,
1180 # just the number of posts in the current results window
1184 $anchor = '#u'; # top of only message
1186 $n = " ($n+ messages)";
1187 $anchor = '#t'; # thread skeleton
1190 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1193 my $level = shift @extra;
1194 my $subj = shift @extra; # already normalized
1195 $mid = delete $seen->{$subj};
1196 my @subj = split(/ /, $subj);
1197 my @next_prev = @subj; # full copy
1198 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1199 $prev_subj = \@next_prev;
1200 $subj = join(' ', @subj);
1201 $subj = ascii_html($subj);
1202 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1203 $href = mid_href($mid);
1204 $s .= indent_for($level) . TCHILD;
1205 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1209 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1214 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1215 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1218 sub pagination_footer ($$) {
1219 my ($ctx, $latest) = @_;
1220 my $next = $ctx->{next_page} || '';
1221 my $prev = $ctx->{prev_page} || '';
1222 if ($prev) { # aligned padding for: 'next (older) | '
1223 $next = $next ? "$next | " : ' | ';
1224 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1226 ($next || $prev) ? "<hr><pre id=nav>page: $next$prev</pre>" : '';
1229 sub paginate_recent ($$) {
1230 my ($ctx, $lim) = @_;
1231 my $t = $ctx->{qp}->{t} || '';
1232 my $opts = { limit => $lim };
1233 my ($after, $before);
1235 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1236 # if only $after exists "YYYYMMDD.." because "." could be skipped
1237 # if interpreted as an end-of-sentence
1238 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1239 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1241 my $msgs = $ctx->{ibx}->over->recent($opts, $after, $before);
1242 if (defined($after) && scalar(@$msgs) < $lim) {
1243 $after = $before = undef;
1244 $msgs = $ctx->{ibx}->over->recent($opts);
1246 my $more = scalar(@$msgs) == $lim;
1247 my ($newest, $oldest);
1249 $newest = $msgs->[0]->{ts};
1250 $oldest = $msgs->[-1]->{ts};
1251 # if we only had $after, our SQL query in ->recent ordered
1252 if ($newest < $oldest) {
1253 ($oldest, $newest) = ($newest, $oldest);
1254 $more = undef if defined($after) && $after < $oldest;
1256 if (defined($after // $before)) {
1257 my $n = strftime('%Y-%m-%d %H:%M:%S', gmtime($newest));
1258 my $o = strftime('%Y-%m-%d %H:%M:%S', gmtime($oldest));
1259 $ctx->{t_note} = <<EOM;
1260 messages from $o to $n UTC, [<a href="#nav">more...</a>]
1262 my $s = ts2str($newest);
1263 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1267 if (defined($oldest) && $more) {
1268 my $s = ts2str($oldest);
1269 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1275 # GET /$INBOX - top-level inbox view for indexed inboxes
1278 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1279 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic) if @$msgs;
1280 html_oneshot($ctx, dump_topics($ctx), pagination_footer($ctx, '.'));
1283 sub thread_adj_level {
1284 my ($ctx, $level) = @_;
1286 my $max = $ctx->{cur_level};
1288 return ('', '') if $max == 0; # flat output
1290 # reset existing lists
1291 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1292 $ctx->{cur_level} = 0;
1294 } elsif ($level == $max) { # continue existing list
1296 } elsif ($level < $max) {
1297 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1298 $ctx->{cur_level} = $level;
1299 ("$beg<li>", '</li>');
1300 } else { # ($level > $max) # start a new level
1301 $ctx->{cur_level} = $level;
1302 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1307 sub ghost_index_entry {
1308 my ($ctx, $level, $node) = @_;
1309 my ($beg, $end) = thread_adj_level($ctx, $level);
1310 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid} // '?')