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 my $cur = delete $ctx->{smsg} or return; # undef: done
33 if (my $over = $ctx->{-inbox}->over) {
34 $nxt = $ctx->{smsg} = $over->next_by_mid(@{$ctx->{next_arg}});
36 $ctx->{mhref} = ($ctx->{nr} || $nxt) ?
37 "../${\mid_href($cur->{mid})}/" : '';
38 my $eml = $ctx->{-inbox}->smsg_eml($cur) or return;
39 my $hdr = $eml->header_obj;
40 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
41 multipart_text_as_html($eml, $ctx);
43 $$obuf .= '</pre><hr>';
44 # we want to at least show the message if something
46 eval { $$obuf .= html_footer($ctx, $ctx->{first_hdr}) } if !$nxt;
50 # /$INBOX/$MESSAGE_ID/ for unindexed v1 inboxes
53 my $eml = delete $ctx->{eml} or return;
54 my $hdr = $eml->header_obj;
56 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
57 multipart_text_as_html($eml, $ctx);
59 $$obuf .= '</pre><hr>';
60 eval { $$obuf .= html_footer($ctx, $hdr) };
64 sub no_over_html ($) {
66 my $bref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return; # 404
67 $ctx->{eml} = PublicInbox::Eml->new($bref);
68 PublicInbox::WwwStream::response($ctx, 200, \&no_over_i);
71 # public functions: (unstable)
75 my $ibx = $ctx->{-inbox};
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 ];
80 $ctx->{smsg} = $over->next_by_mid(@$next_arg) or return;
81 PublicInbox::WwwStream::response($ctx, 200, \&msg_page_i);
84 # /$INBOX/$MESSAGE_ID/#R
88 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
90 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
93 my $ibx = $ctx->{-inbox};
94 if (my $url = $ibx->{infourl}) {
95 $url = prurl($ctx->{env}, $url);
96 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
99 my ($arg, $link, $reply_to_all) =
100 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
101 if (ref($arg) eq 'SCALAR') {
102 return '<pre id=R>'.ascii_html($$arg).'</pre>';
105 # mailto: link only works if address obfuscation is disabled
109 * If your mail client supports setting the <b>In-Reply-To</b> header
110 via mailto: links, try the <a
111 href="$link">mailto: link</a>
115 push @$arg, '/path/to/YOUR_REPLY';
116 $arg = ascii_html(join(" \\\n ", '', @$arg));
119 id=R><b>Reply instructions:</b>
121 You may reply publicly to <a
122 href=#t>this message</a> via plain-text email
123 using any one of the following methods:
125 * Save the following mbox file, import it into your mail client,
126 and $reply_to_all from there: <a
129 Avoid top-posting and favor interleaved quoting:
131 href="$p_url">$p_url</a>
133 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
134 switches of git-send-email(1):
139 href="$se_url">$se_url</a>
146 my $refs = references($hdr);
150 sub fold_addresses ($) {
151 return $_[0] if length($_[0]) <= COLS;
152 # try to fold on commas after non-word chars before $lim chars,
153 # Try to get the "," preceded by ">" or ")", but avoid folding
154 # on the comma where somebody uses "Lastname, Firstname".
155 # We also try to keep the last and penultimate addresses in
156 # the list on the same line if possible, hence the extra \z
157 # Fall back to folding on spaces at $lim + 1 chars
158 my $lim = COLS - 8; # 8 = "\t" display width
159 my $too_long = $lim + 1;
160 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
162 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
165 .{$too_long,}?)(?:\s|\z)/xgo));
168 sub _hdr_names_html ($$) {
169 my ($hdr, $field) = @_;
170 my @vals = $hdr->header($field) or return '';
171 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
175 my ($nr, $singular, $plural) = @_;
176 return "0 $plural" if $nr == 0;
177 $nr == 1 ? "$nr $singular" : "$nr $plural";
180 # human-friendly format
181 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
183 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
184 # this is already inside a <pre>
186 my ($ctx, $smsg, $eml, $more) = @_;
187 my $subj = delete $smsg->{subject};
188 my $mid_raw = $smsg->{mid};
189 my $id = id_compress($mid_raw, 1);
191 my $root_anchor = $ctx->{root_anchor} || '';
193 my $obfs_ibx = $ctx->{-obfs_ibx};
195 $subj = '(no subject)' if $subj eq '';
196 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
197 $subj = '<b>'.ascii_html($subj).'</b>';
198 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
199 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
201 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
203 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
205 # Deleting these fields saves about 400K as we iterate across 1K msgs
206 delete @$smsg{qw(ts blob)};
208 my $hdr = $eml->header_obj;
209 my $from = _hdr_names_html($hdr, 'From');
210 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
211 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
212 my $upfx = $ctx->{-upfx};
213 my $mhref = $upfx . mid_href($mid_raw) . '/';
214 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
215 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
216 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
217 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
218 my ($tlen, $clen) = (length($to), length($cc));
220 if (($tlen + $clen) > COLS) {
221 $to_cc .= ' To: '.$to."\n" if $tlen;
222 $to_cc .= ' Cc: '.$cc."\n" if $clen;
225 $to_cc .= ' To: '.$to;
226 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
228 $to_cc .= ' Cc: '.$cc if $clen;
232 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
235 my $mapping = $ctx->{mapping};
236 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
237 my $href = $upfx . mid_href($irt) . '/';
238 my $html = ascii_html($irt);
239 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
243 # scan through all parts, looking for displayable text
244 $ctx->{mhref} = $mhref;
246 $eml->each_part(\&add_text_body, $ctx, 1);
250 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
251 "<a\nhref=\"$mhref\">permalink</a>" .
252 " <a\nhref=\"${mhref}raw\">raw</a>" .
253 " <a\nhref=\"${mhref}#R\">reply</a>";
256 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
257 $rv .= "\t[relevance $pct%]";
260 my $nested = 'nested';
265 $flat = "<b>$flat</b>";
267 $nested = "<b>$nested</b>";
269 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
270 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
271 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
276 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
280 sub pad_link ($$;$) {
281 my ($mid, $level, $s) = @_;
283 my $id = id_compress($mid, 1);
284 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
288 my ($mid_raw, $irt, $id, $ctx) = @_;
290 my $mapping = $ctx->{mapping} or return $rv;
292 my $mid_map = $mapping->{$mid_raw};
294 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
295 my ($attr, $node, $idx, $level) = @$mid_map;
296 my $children = $node->{children};
297 my $nr_c = scalar @$children;
300 # delete saves about 200KB on a 1K message thread
301 if (my $refs = delete $node->{references}) {
302 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
304 my $irt_map = $mapping->{$$irt} if defined $$irt;
305 if (defined $irt_map) {
306 $siblings = $irt_map->[1]->{children};
307 $nr_s = scalar(@$siblings) - 1;
308 $rv .= $pad . $irt_map->[0];
310 my $prev = $siblings->[$idx - 1];
311 my $pmid = $prev->{mid};
313 my $s = ($idx - 1). ' preceding siblings ...';
314 $rv .= pad_link($pmid, $level, $s);
315 } elsif ($idx == 2) {
316 my $ppmid = $siblings->[0]->{mid};
317 $rv .= $pad . $mapping->{$ppmid}->[0];
319 $rv .= $pad . $mapping->{$pmid}->[0];
322 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
323 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
324 $attr =~ s!\n\z!</b>\n!s;
325 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
326 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
329 my $cmid = $children->[0]->{mid};
330 $rv .= $pad . $mapping->{$cmid}->[0];
332 my $s = ($nr_c - 1). ' more replies';
333 $rv .= pad_link($cmid, $level + 1, $s);
334 } elsif (my $cn = $children->[1]) {
335 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
339 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
341 my $nmid = $next->{mid};
342 $rv .= $pad . $mapping->{$nmid}->[0];
343 my $nnext = $nr_s - $idx;
345 my $s = ($nnext - 1).' subsequent siblings';
346 $rv .= pad_link($nmid, $level, $s);
347 } elsif (my $nn = $siblings->[$idx + 2]) {
348 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
351 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
354 # non-recursive thread walker
355 sub walk_thread ($$$) {
356 my ($rootset, $ctx, $cb) = @_;
357 my @q = map { (0, $_, -1) } @$rootset;
359 my ($level, $node, $i) = splice(@q, 0, 3);
360 defined $node or next;
361 $cb->($ctx, $level, $node, $i) or return;
364 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
368 sub pre_thread { # walk_thread callback
369 my ($ctx, $level, $node, $idx) = @_;
370 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
371 skel_dump($ctx, $level, $node);
374 sub thread_eml_entry {
375 my ($ctx, $level, $smsg, $eml) = @_;
376 my ($beg, $end) = thread_adj_level($ctx, $level);
377 $beg . '<pre>' . eml_entry($ctx, $smsg, $eml, 0) . '</pre>' . $end;
380 sub next_in_queue ($;$) {
381 my ($q, $ghost_ok) = @_;
383 my ($level, $smsg) = splice(@$q, 0, 2);
385 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
386 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
391 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
393 return unless exists($ctx->{skel});
394 my $nr = $ctx->{nr}++;
395 my ($level, $smsg) = next_in_queue($ctx->{-queue}, $nr);
398 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{skel}};
400 my $eml = $ctx->{-inbox}->smsg_eml($smsg) or return
401 ghost_index_entry($ctx, $level, $smsg);
404 $ctx->{-title_html} = ascii_html($smsg->{subject});
405 $ctx->html_top . thread_eml_entry($ctx, $level, $smsg, $eml);
407 thread_eml_entry($ctx, $level, $smsg, $eml);
411 sub stream_thread ($$) {
412 my ($rootset, $ctx) = @_;
413 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
414 PublicInbox::WwwStream::response($ctx, 200, \&stream_thread_i);
417 # /$INBOX/$MESSAGE_ID/t/
420 my $mid = $ctx->{mid};
421 my $ibx = $ctx->{-inbox};
422 my ($nr, $msgs) = $ibx->over->get_thread($mid);
423 return missing_thread($ctx) if $nr == 0;
424 my $skel = '<hr><pre>';
425 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
426 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
427 $skel .= "<b\nid=t>Thread overview:</b> ";
428 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
429 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
430 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
431 $skel .= "-- links below jump to the message on this page --\n";
432 $ctx->{-upfx} = '../../';
433 $ctx->{cur_level} = 0;
434 $ctx->{skel} = \$skel;
435 $ctx->{prev_attr} = '';
436 $ctx->{prev_level} = 0;
437 $ctx->{root_anchor} = anchor_for($mid);
438 $ctx->{mapping} = {};
439 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
442 my $rootset = thread_results($ctx, $msgs);
444 # reduce hash lookups in pre_thread->skel_dump
445 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
446 walk_thread($rootset, $ctx, \&pre_thread);
449 return stream_thread($rootset, $ctx) unless $ctx->{flat};
451 # flat display: lazy load the full message from smsg
452 $ctx->{msgs} = $msgs;
453 $ctx->{-html_tip} = '<pre>';
454 PublicInbox::WwwStream::response($ctx, 200, \&thread_html_i);
457 sub thread_html_i { # PublicInbox::WwwStream::getline callback
459 my $msgs = $ctx->{msgs} or return;
460 while (my $smsg = shift @$msgs) {
461 my $eml = $ctx->{-inbox}->smsg_eml($smsg) or next;
462 if (exists $ctx->{-html_tip}) {
463 $ctx->{-title_html} = ascii_html($smsg->{subject});
464 return $ctx->html_top .
465 eml_entry($ctx, $smsg, $eml, scalar @$msgs);
467 return eml_entry($ctx, $smsg, $eml, scalar @$msgs);
469 my ($skel) = delete @$ctx{qw(skel msgs)};
473 sub multipart_text_as_html {
474 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
476 # scan through all parts, looking for displayable text
477 $_[0]->each_part(\&add_text_body, $_[1], 1);
480 sub submsg_hdr ($$) {
481 my ($ctx, $eml) = @_;
482 my $obfs_ibx = $ctx->{-obfs_ibx};
483 my $rv = $ctx->{obuf};
485 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
486 my @v = $eml->header($h);
488 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
495 sub attach_link ($$$$;$) {
496 my ($ctx, $ct, $p, $fn, $err) = @_;
497 my ($part, $depth, $idx) = @$p;
499 # Eml iteration clobbers multipart ->{bdy}, so do not offer
500 # downloads for 0-byte multipart attachments
501 return unless $part->{bdy};
503 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
504 my $size = bytes::length($part->body);
506 # hide attributes normally, unless we want to aid users in
507 # spotting MUA problems:
508 $ct =~ s/;.*// unless $err;
509 $ct = ascii_html($ct);
511 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
513 } elsif ($ct eq 'text/plain') {
518 my $rv = $ctx->{obuf};
519 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
522 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
525 $$rv .= "[-- Attachment #$idx: ";
526 my $ts = "Type: $ct, Size: $size bytes";
527 my $desc = $part->header('Content-Description') // $fn // '';
528 $desc = ascii_html($desc);
529 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
532 submsg_hdr($ctx, $part) if $part->{is_submsg};
537 sub add_text_body { # callback for each_part
539 my $upfx = $ctx->{mhref};
540 my $ibx = $ctx->{-inbox};
541 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
542 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
543 my ($part, $depth, $idx) = @$p;
544 my $ct = $part->content_type || 'text/plain';
545 my $fn = $part->filename;
546 my ($s, $err) = msg_part_text($part, $ct);
547 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
549 my $rv = $ctx->{obuf};
550 if ($part->{is_submsg}) {
551 submsg_hdr($ctx, $part);
555 # makes no difference to browsers, and don't screw up filename
556 # link generation in diffs with the extra '%0D'
559 # will be escaped to `•' in HTML
560 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
562 # always support diff-highlighting, but we can't linkify hunk
563 # headers for solver unless some coderepo are configured:
565 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
566 # diffstat anchors do not link across attachments or messages,
567 # -apfx is just a stable prefix for making diffstat anchors
568 # linkable to the first diff hunk w/o crossing attachments
569 $idx =~ tr!.!/!; # compatibility with previous versions
570 $ctx->{-apfx} = $upfx . $idx;
572 # do attr => filename mappings for diffstats in git diffs:
573 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
575 delete $ctx->{-long_path};
577 if ($ibx->{-repo_objs}) {
578 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
580 $spfx =~ s!/([^/]*)/\z!/!;
582 my $n_slash = $upfx =~ tr!/!/!;
585 } elsif ($n_slash == 1) {
587 } else { # nslash == 2
592 $ctx->{-spfx} = $spfx;
595 # some editors don't put trailing newlines at the end:
596 $s .= "\n" unless $s =~ /\n\z/s;
598 # split off quoted and unquoted blocks:
599 my @sections = PublicInbox::MsgIter::split_quotes($s);
600 undef $s; # free memory
601 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
602 # badly-encoded message with $err? tell the world about it!
603 attach_link($ctx, $ct, $p, $fn, $err);
606 foreach my $cur (@sections) {
608 # we use a <span> here to allow users to specify
609 # their own color for quoted text
610 $$rv .= qq(<span\nclass="q">);
611 $$rv .= $l->to_html($cur);
614 flush_diff($ctx, \$cur);
617 $$rv .= $l->to_html($cur);
619 undef $cur; # free memory
623 sub _msg_page_prepare_obuf {
624 my ($hdr, $ctx) = @_;
625 my $over = $ctx->{-inbox}->over;
626 my $obfs_ibx = $ctx->{-obfs_ibx};
628 my $mids = mids_for_index($hdr);
629 my $nr = $ctx->{nr}++;
630 if ($nr) { # unlikely
633 $ctx->{first_hdr} = $hdr;
636 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
638 $rv .= "<pre\nid=b>"; # anchor for body start
640 $ctx->{-upfx} = '../' if $over;
641 my @title; # (Subject[0], From[0])
642 for my $v ($hdr->header('From')) {
643 my @n = PublicInbox::Address::names($v);
645 $title[1] //= ascii_html(join(', ', @n));
647 obfuscate_addrs($obfs_ibx, $v);
648 obfuscate_addrs($obfs_ibx, $title[1]);
650 $rv .= "From: $v\n" if $v ne '';
652 foreach my $h (qw(To Cc)) {
653 for my $v ($hdr->header($h)) {
656 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
657 $rv .= "$h: $v\n" if $v ne '';
660 my @subj = $hdr->header('Subject');
662 my $v = ascii_html(shift @subj);
663 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
665 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
667 for $v (@subj) { # multi-Subject message :<
669 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
670 $rv .= "Subject: $v\n";
672 } else { # dummy anchor for thread skeleton at bottom of page
673 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
674 $title[0] = '(no subject)';
676 for my $v ($hdr->header('Date')) {
678 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
681 if (!$nr) { # first (and only) message, common case
682 $ctx->{-title_html} = join(' - ', @title);
683 $rv = $ctx->html_top . $rv;
685 if (scalar(@$mids) == 1) { # common case
686 my $mhtml = ascii_html($mids->[0]);
687 $rv .= "Message-ID: <$mhtml> ";
688 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
690 # X-Alt-Message-ID can happen if a message is injected from
691 # public-inbox-nntpd because of multiple Message-ID headers.
692 my $lnk = PublicInbox::Linkify->new;
694 for my $h (qw(Message-ID X-Alt-Message-ID)) {
695 $s .= "$h: $_\n" for ($hdr->header_raw($h));
697 $lnk->linkify_mids('..', \$s, 1);
700 $rv .= _parent_headers($hdr, $over);
706 qq(expand[<a\nhref="T/#u">flat</a>) .
707 qq(|<a\nhref="t/#u">nested</a>] ) .
708 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
709 qq(<a\nhref="t.atom">Atom feed</a>);
712 sub thread_skel ($$$) {
713 my ($skel, $ctx, $hdr) = @_;
714 my $mid = mids($hdr)->[0];
715 my $ibx = $ctx->{-inbox};
716 my ($nr, $msgs) = $ibx->over->get_thread($mid);
717 my $parent = in_reply_to($hdr);
718 $$skel .= "\n<b>Thread overview: </b>";
720 if (defined $parent) {
721 $$skel .= SKEL_EXPAND."\n ";
722 $$skel .= ghost_parent('../', $parent) . "\n";
724 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
726 $ctx->{next_msg} = undef;
727 $ctx->{parent_msg} = $parent;
732 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
734 # nb: mutt only shows the first Subject in the index pane
735 # when multiple Subject: headers are present, so we follow suit:
736 my $subj = $hdr->header('Subject') // '';
737 $subj = '(no subject)' if $subj eq '';
738 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
740 $ctx->{prev_attr} = '';
741 $ctx->{prev_level} = 0;
742 $ctx->{skel} = $skel;
744 # reduce hash lookups in skel_dump
745 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
746 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
748 $ctx->{parent_msg} = $parent;
751 sub _parent_headers {
752 my ($hdr, $over) = @_;
754 my @irt = $hdr->header_raw('In-Reply-To');
757 my $lnk = PublicInbox::Linkify->new;
758 $rv .= "In-Reply-To: $_\n" for @irt;
759 $lnk->linkify_mids('..', \$rv);
761 $refs = references($hdr);
762 my $irt = pop @$refs;
764 my $html = ascii_html($irt);
765 my $href = mid_href($irt);
766 $rv .= "In-Reply-To: <";
767 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
771 # do not display References: if search is present,
772 # we show the thread skeleton at the bottom, instead.
775 $refs //= references($hdr);
777 @$refs = map { linkify_ref_no_over($_) } @$refs;
778 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
783 # returns a string buffer via ->getline
785 my ($ctx, $hdr) = @_;
786 my $ibx = $ctx->{-inbox};
788 my $skel = " <a\nhref=\"$upfx\">index</a>";
792 thread_skel(\$skel, $ctx, $hdr);
797 if (my $n = $ctx->{next_msg}) {
799 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
802 my $par = $ctx->{parent_msg};
807 if (my $p = $ctx->{prev_msg}) {
808 $prev = mid_href($p);
809 if ($p && $par && $p eq $par) {
810 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
811 'rel=prev>prev parent</a>';
814 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
816 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
818 } elsif ($u) { # unlikely
819 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
821 $rv .= "$next $prev$parent ";
823 $rv .= qq(<a\nhref="#R">reply</a>);
826 $rv .= msg_reply($ctx, $hdr);
829 sub linkify_ref_no_over {
831 my $href = mid_href($mid);
832 my $html = ascii_html($mid);
833 "<<a\nhref=\"../$href/\">$html</a>>";
838 'm' . id_compress($msgid, 1);
842 my ($upfx, $mid) = @_;
844 my $href = mid_href($mid);
845 my $html = ascii_html($mid);
846 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
851 $level ? INDENT x ($level - 1) : '';
855 my ($ctx, $level, $node, $idx) = @_;
856 ++$ctx->{root_idx} if $level == 0;
857 if ($node->{mid} eq $ctx->{mid}) {
858 $ctx->{found_mid_at} = $ctx->{root_idx};
864 sub strict_loose_note ($) {
867 " -- strict thread matches above, loose matches on Subject: below --\n";
869 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
871 " -- use mbox.gz link to download all $nr messages --\n";
877 my ($ctx, $msgs) = @_;
878 require PublicInbox::SearchThread;
879 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
881 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
882 # and preserved in the future. This bug is hidden by `sid' matches
883 # in get_thread, so we never noticed it until now. And even when
884 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
885 # bump since reindexing is expensive and users may not do it
887 # loose threading could've returned too many results,
888 # put the root the message we care about at the top:
889 my $mid = $ctx->{mid};
890 if (defined($mid) && scalar(@$rootset) > 1) {
891 $ctx->{root_idx} = -1;
892 my $nr = scalar @$msgs;
893 walk_thread($rootset, $ctx, \&find_mid_root);
894 my $idx = $ctx->{found_mid_at};
895 if (defined($idx) && $idx != 0) {
896 my $tip = splice(@$rootset, $idx, 1);
897 @$rootset = reverse @$rootset;
898 unshift @$rootset, $tip;
899 $ctx->{sl_note} = strict_loose_note($nr);
907 require PublicInbox::ExtMsg;
908 PublicInbox::ExtMsg::ext_msg($ctx);
912 my ($prev_subj, $subj, $val) = @_;
914 my $omit; # '"' denotes identical text omitted
915 my (@prev_pop, @curr_pop);
916 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
917 push(@prev_pop, pop(@$prev_subj));
918 push(@curr_pop, pop(@$subj));
921 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
922 if (scalar(@curr_pop) == 1) {
924 push @$prev_subj, @prev_pop;
925 push @$subj, @curr_pop;
930 sub skel_dump { # walk_thread callback
931 my ($ctx, $level, $smsg) = @_;
932 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
934 my $skel = $ctx->{skel};
935 my $cur = $ctx->{cur};
936 my $mid = $smsg->{mid};
938 if ($level == 0 && $ctx->{skel_dump_roots}++) {
939 $$skel .= delete($ctx->{sl_note}) || '';
942 my $f = ascii_html($smsg->{from_name});
943 my $obfs_ibx = $ctx->{-obfs_ibx};
944 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
946 my $d = fmt_ts($smsg->{ds});
947 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
948 if (exists $ctx->{searchview}) {
949 if (defined(my $pct = $smsg->{pct})) {
950 $d .= (sprintf(' % 2u', $pct) . '%');
956 $d .= ' ' . indent_for($level) . th_pfx($level);
958 $ctx->{first_level} ||= $level;
960 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
961 $ctx->{prev_attr} = $attr;
963 $ctx->{prev_level} = $level;
968 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
969 "$attr [this message]</a></b>\n";
972 $ctx->{prev_msg} = $mid;
975 $ctx->{next_msg} ||= $mid;
978 # Subject is never undef, this mail was loaded from
979 # our Xapian which would've resulted in '' if it were
980 # really missing (and Filter rejects empty subjects)
981 my @subj = split(/ /, subject_normalized($smsg->{subject}));
982 # remove common suffixes from the subject if it matches the previous,
983 # so we do not show redundant text at the end.
984 my $prev_subj = $ctx->{prev_subj} || [];
985 $ctx->{prev_subj} = [ @subj ];
986 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
989 my $subj = join(' ', @subj);
990 $subj = ascii_html($subj);
991 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
992 $end = "$subj</a> $omit$f\n"
998 my $mapping = $unmatched ? undef : $ctx->{mapping};
1000 my $map = $mapping->{$mid};
1001 $id = id_compress($mid, 1);
1003 $map->[0] = "$d<a\nhref=\"$m\">$end";
1006 $m = $ctx->{-upfx}.mid_href($mid).'/';
1008 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1013 my ($ctx, $level, $node) = @_;
1015 my $mid = $node->{mid};
1016 my $d = ' [not found] ';
1017 $d .= ' ' if exists $ctx->{searchview};
1018 $d .= indent_for($level) . th_pfx($level);
1019 my $upfx = $ctx->{-upfx};
1020 my $href = $upfx . mid_href($mid) . '/';
1021 my $html = ascii_html($mid);
1023 my $mapping = $ctx->{mapping};
1024 my $map = $mapping->{$mid} if $mapping;
1026 my $id = id_compress($mid, 1);
1027 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1028 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1030 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1032 ${$ctx->{skel}} .= $d;
1038 (eval { $a->topmost->{ds} } || 0) <=>
1039 (eval { $b->topmost->{ds} } || 0)
1043 # accumulate recent topics if search is supported
1044 # returns 200 if done, 404 if not
1045 sub acc_topic { # walk_thread callback
1046 my ($ctx, $level, $smsg) = @_;
1047 my $mid = $smsg->{mid};
1048 my $has_blob = $smsg->{blob} // do {
1049 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1050 %$smsg = (%$smsg, %$by_mid);
1055 my $subj = subject_normalized($smsg->{subject});
1056 $subj = '(no subject)' if $subj eq '';
1057 my $ds = $smsg->{ds};
1058 if ($level == 0) { # new, top-level topic
1059 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1060 $ctx->{-cur_topic} = $topic;
1061 push @{$ctx->{order}}, $topic;
1065 # continue existing topic
1066 my $topic = $ctx->{-cur_topic}; # should never be undef
1067 $topic->[0] = $ds if $ds > $topic->[0];
1068 $topic->[1]++; # bump N+ message counter
1069 my $seen = $topic->[2];
1070 if (scalar(@$topic) == 3) { # parent was a ghost
1071 push @$topic, $subj;
1072 } elsif (!defined($seen->{$subj})) {
1073 push @$topic, $level, $subj; # @extra messages
1075 $seen->{$subj} = $mid; # latest for subject
1076 } else { # ghost message
1077 return 1 if $level != 0; # ignore child ghosts
1078 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1079 push @{$ctx->{order}}, $topic;
1086 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1088 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1093 my $ibx = $ctx->{-inbox};
1094 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1096 # sort by recency, this allows new posts to "bump" old topics...
1097 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1098 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1100 next unless defined $top_subj; # ghost topic
1101 my $mid = delete $seen->{$top_subj};
1102 my $href = mid_href($mid);
1103 my $prev_subj = [ split(/ /, $top_subj) ];
1104 $top_subj = ascii_html($top_subj);
1107 # $n isn't the total number of posts on the topic,
1108 # just the number of posts in the current results window
1112 $anchor = '#u'; # top of only message
1114 $n = " ($n+ messages)";
1115 $anchor = '#t'; # thread skeleton
1118 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1119 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1120 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1121 " $ds UTC $n - $mbox / $atom\n";
1122 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1123 my $level = $extra[$i];
1124 my $subj = $extra[$i + 1]; # already normalized
1125 $mid = delete $seen->{$subj};
1126 my @subj = split(/ /, $subj);
1127 my @next_prev = @subj; # full copy
1128 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1129 $prev_subj = \@next_prev;
1130 $subj = join(' ', @subj);
1131 $subj = ascii_html($subj);
1132 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1133 $href = mid_href($mid);
1134 $s .= indent_for($level) . TCHILD;
1135 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1139 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1143 # only for the t= query parameter passed to overview DB
1144 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1147 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1148 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1151 sub pagination_footer ($$) {
1152 my ($ctx, $latest) = @_;
1153 delete $ctx->{qp} or return;
1154 my $next = $ctx->{next_page} || '';
1155 my $prev = $ctx->{prev_page} || '';
1157 $next = $next ? "$next " : ' ';
1158 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1160 "<hr><pre>page: $next$prev</pre>";
1163 sub index_nav { # callback for WwwStream::getline
1165 return $ctx->html_top if exists $ctx->{-html_tip};
1166 pagination_footer($ctx, '.')
1169 sub paginate_recent ($$) {
1170 my ($ctx, $lim) = @_;
1171 my $t = $ctx->{qp}->{t} || '';
1172 my $opts = { limit => $lim };
1173 my ($after, $before);
1175 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1176 # if only $after exists "YYYYMMDD.." because "." could be skipped
1177 # if interpreted as an end-of-sentence
1178 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1179 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1181 my $ibx = $ctx->{-inbox};
1182 my $msgs = $ibx->recent($opts, $after, $before);
1183 my $nr = scalar @$msgs;
1184 if ($nr < $lim && defined($after)) {
1185 $after = $before = undef;
1186 $msgs = $ibx->recent($opts);
1187 $nr = scalar @$msgs;
1189 my $more = $nr == $lim;
1190 my ($newest, $oldest);
1192 $newest = $msgs->[0]->{ts};
1193 $oldest = $msgs->[-1]->{ts};
1194 # if we only had $after, our SQL query in ->recent ordered
1195 if ($newest < $oldest) {
1196 ($oldest, $newest) = ($newest, $oldest);
1197 $more = 0 if defined($after) && $after < $oldest;
1200 if (defined($oldest) && $more) {
1201 my $s = ts2str($oldest);
1202 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1204 if (defined($newest) && (defined($before) || defined($after))) {
1205 my $s = ts2str($newest);
1206 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1213 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1215 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1217 PublicInbox::WwwStream::response($ctx, dump_topics($ctx), \&index_nav);
1220 sub thread_adj_level {
1221 my ($ctx, $level) = @_;
1223 my $max = $ctx->{cur_level};
1225 return ('', '') if $max == 0; # flat output
1227 # reset existing lists
1228 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1229 $ctx->{cur_level} = 0;
1231 } elsif ($level == $max) { # continue existing list
1233 } elsif ($level < $max) {
1234 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1235 $ctx->{cur_level} = $level;
1236 ("$beg<li>", '</li>');
1237 } else { # ($level > $max) # start a new level
1238 $ctx->{cur_level} = $level;
1239 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1244 sub ghost_index_entry {
1245 my ($ctx, $level, $node) = @_;
1246 my ($beg, $end) = thread_adj_level($ctx, $level);
1247 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})