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 qw(html_oneshot);
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 $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
37 multipart_text_as_html($eml, $ctx);
39 $$obuf .= '</pre><hr>';
40 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
42 } else { # called by WwwStream::async_next or getline
43 $ctx->{smsg}; # may be undef
47 # /$INBOX/$MSGID/ for unindexed v1 inboxes
48 sub no_over_html ($) {
50 my $bref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return; # 404
51 my $eml = PublicInbox::Eml->new($bref);
53 PublicInbox::WwwStream::init($ctx);
54 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
55 multipart_text_as_html($eml, $ctx);
57 $$obuf .= '</pre><hr>';
58 eval { $$obuf .= html_footer($ctx, $eml) };
59 html_oneshot($ctx, 200, $obuf);
62 # public functions: (unstable)
66 my $ibx = $ctx->{-inbox};
67 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
68 my $over = $ctx->{over} = $ibx->over or return no_over_html($ctx);
70 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
71 $ctx->{smsg} = $over->next_by_mid(@$next_arg) or return; # undef == 404
72 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
75 # /$INBOX/$MESSAGE_ID/#R
79 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
81 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
84 my $ibx = $ctx->{-inbox};
85 if (my $url = $ibx->{infourl}) {
86 $url = prurl($ctx->{env}, $url);
87 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
90 my ($arg, $link, $reply_to_all) =
91 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
92 if (ref($arg) eq 'SCALAR') {
93 return '<pre id=R>'.ascii_html($$arg).'</pre>';
96 # mailto: link only works if address obfuscation is disabled
100 * If your mail client supports setting the <b>In-Reply-To</b> header
101 via mailto: links, try the <a
102 href="$link">mailto: link</a>
106 push @$arg, '/path/to/YOUR_REPLY';
107 $arg = ascii_html(join(" \\\n ", '', @$arg));
110 id=R><b>Reply instructions:</b>
112 You may reply publicly to <a
113 href=#t>this message</a> via plain-text email
114 using any one of the following methods:
116 * Save the following mbox file, import it into your mail client,
117 and $reply_to_all from there: <a
120 Avoid top-posting and favor interleaved quoting:
122 href="$p_url">$p_url</a>
124 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
125 switches of git-send-email(1):
130 href="$se_url">$se_url</a>
137 my $refs = references($hdr);
141 sub fold_addresses ($) {
142 return $_[0] if length($_[0]) <= COLS;
143 # try to fold on commas after non-word chars before $lim chars,
144 # Try to get the "," preceded by ">" or ")", but avoid folding
145 # on the comma where somebody uses "Lastname, Firstname".
146 # We also try to keep the last and penultimate addresses in
147 # the list on the same line if possible, hence the extra \z
148 # Fall back to folding on spaces at $lim + 1 chars
149 my $lim = COLS - 8; # 8 = "\t" display width
150 my $too_long = $lim + 1;
151 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
153 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
156 .{$too_long,}?)(?:\s|\z)/xgo));
159 sub _hdr_names_html ($$) {
160 my ($hdr, $field) = @_;
161 my @vals = $hdr->header($field) or return '';
162 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
166 my ($nr, $singular, $plural) = @_;
167 return "0 $plural" if $nr == 0;
168 $nr == 1 ? "$nr $singular" : "$nr $plural";
171 # human-friendly format
172 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
174 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
175 # this is already inside a <pre>
177 my ($ctx, $eml) = @_;
178 my $smsg = delete $ctx->{smsg};
179 my $subj = delete $smsg->{subject};
180 my $mid_raw = $smsg->{mid};
181 my $id = id_compress($mid_raw, 1);
183 my $root_anchor = $ctx->{root_anchor} || '';
185 my $obfs_ibx = $ctx->{-obfs_ibx};
187 $subj = '(no subject)' if $subj eq '';
188 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
189 $subj = '<b>'.ascii_html($subj).'</b>';
190 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
191 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
193 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
195 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
197 # Deleting these fields saves about 400K as we iterate across 1K msgs
198 delete @$smsg{qw(ts blob)};
200 my $from = _hdr_names_html($eml, 'From');
201 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
202 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
203 my $upfx = $ctx->{-upfx};
204 my $mhref = $upfx . mid_href($mid_raw) . '/';
205 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
206 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
207 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
208 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
209 my ($tlen, $clen) = (length($to), length($cc));
211 if (($tlen + $clen) > COLS) {
212 $to_cc .= ' To: '.$to."\n" if $tlen;
213 $to_cc .= ' Cc: '.$cc."\n" if $clen;
216 $to_cc .= ' To: '.$to;
217 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
219 $to_cc .= ' Cc: '.$cc if $clen;
223 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
226 my $mapping = $ctx->{mapping};
227 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
228 my $href = $upfx . mid_href($irt) . '/';
229 my $html = ascii_html($irt);
230 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
234 # scan through all parts, looking for displayable text
235 $ctx->{mhref} = $mhref;
237 $eml->each_part(\&add_text_body, $ctx, 1);
241 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
242 "<a\nhref=\"$mhref\">permalink</a>" .
243 " <a\nhref=\"${mhref}raw\">raw</a>" .
244 " <a\nhref=\"${mhref}#R\">reply</a>";
247 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
248 $rv .= "\t[relevance $pct%]";
251 my $nested = 'nested';
256 $flat = "<b>$flat</b>";
258 $nested = "<b>$nested</b>";
260 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
261 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
262 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
267 # do we have more messages? start a new <pre> if so
268 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
272 sub pad_link ($$;$) {
273 my ($mid, $level, $s) = @_;
275 my $id = id_compress($mid, 1);
276 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
280 my ($mid_raw, $irt, $id, $ctx) = @_;
282 my $mapping = $ctx->{mapping} or return $rv;
284 my $mid_map = $mapping->{$mid_raw};
286 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
287 my ($attr, $node, $idx, $level) = @$mid_map;
288 my $children = $node->{children};
289 my $nr_c = scalar @$children;
292 # delete saves about 200KB on a 1K message thread
293 if (my $refs = delete $node->{references}) {
294 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
296 my $irt_map = $mapping->{$$irt} if defined $$irt;
297 if (defined $irt_map) {
298 $siblings = $irt_map->[1]->{children};
299 $nr_s = scalar(@$siblings) - 1;
300 $rv .= $pad . $irt_map->[0];
302 my $prev = $siblings->[$idx - 1];
303 my $pmid = $prev->{mid};
305 my $s = ($idx - 1). ' preceding siblings ...';
306 $rv .= pad_link($pmid, $level, $s);
307 } elsif ($idx == 2) {
308 my $ppmid = $siblings->[0]->{mid};
309 $rv .= $pad . $mapping->{$ppmid}->[0];
311 $rv .= $pad . $mapping->{$pmid}->[0];
314 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
315 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
316 $attr =~ s!\n\z!</b>\n!s;
317 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
318 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
321 my $cmid = $children->[0]->{mid};
322 $rv .= $pad . $mapping->{$cmid}->[0];
324 my $s = ($nr_c - 1). ' more replies';
325 $rv .= pad_link($cmid, $level + 1, $s);
326 } elsif (my $cn = $children->[1]) {
327 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
331 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
333 my $nmid = $next->{mid};
334 $rv .= $pad . $mapping->{$nmid}->[0];
335 my $nnext = $nr_s - $idx;
337 my $s = ($nnext - 1).' subsequent siblings';
338 $rv .= pad_link($nmid, $level, $s);
339 } elsif (my $nn = $siblings->[$idx + 2]) {
340 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
343 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
346 # non-recursive thread walker
347 sub walk_thread ($$$) {
348 my ($rootset, $ctx, $cb) = @_;
349 my @q = map { (0, $_, -1) } @$rootset;
351 my ($level, $node, $i) = splice(@q, 0, 3);
352 defined $node or next;
353 $cb->($ctx, $level, $node, $i) or return;
356 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
360 sub pre_thread { # walk_thread callback
361 my ($ctx, $level, $node, $idx) = @_;
362 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
363 skel_dump($ctx, $level, $node);
366 sub thread_eml_entry {
367 my ($ctx, $eml) = @_;
368 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
369 $beg . '<pre>' . eml_entry($ctx, $eml) . '</pre>' . $end;
372 sub next_in_queue ($$) {
373 my ($q, $ghost_ok) = @_;
375 my ($level, $smsg) = splice(@$q, 0, 2);
377 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
378 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
383 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
384 my ($ctx, $eml) = @_;
385 goto &thread_eml_entry if $eml; # tail recursion
386 return unless exists($ctx->{skel});
387 my $ghost_ok = $ctx->{nr}++;
389 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
391 if (exists $smsg->{blob}) { # next message for cat-file
392 $ctx->{level} = $lvl;
393 if (!$ghost_ok) { # first non-ghost
394 $ctx->{-title_html} =
395 ascii_html($smsg->{subject});
396 $ctx->zmore($ctx->html_top);
400 # buffer the ghost entry and loop
401 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
403 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
404 $ctx->zmore(${delete($ctx->{skel})});
410 sub stream_thread ($$) {
411 my ($rootset, $ctx) = @_;
412 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
413 PublicInbox::WwwStream::aresponse($ctx, 200, \&stream_thread_i);
416 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
419 my $mid = $ctx->{mid};
420 my $ibx = $ctx->{-inbox};
421 my ($nr, $msgs) = $ibx->over->get_thread($mid);
422 return missing_thread($ctx) if $nr == 0;
423 my $skel = '<hr><pre>';
424 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
425 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
426 $skel .= "<b\nid=t>Thread overview:</b> ";
427 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
428 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
429 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
430 $skel .= "-- links below jump to the message on this page --\n";
431 $ctx->{-upfx} = '../../';
432 $ctx->{cur_level} = 0;
433 $ctx->{skel} = \$skel;
434 $ctx->{prev_attr} = '';
435 $ctx->{prev_level} = 0;
436 $ctx->{root_anchor} = anchor_for($mid);
437 $ctx->{mapping} = {};
438 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
441 my $rootset = thread_results($ctx, $msgs);
443 # reduce hash lookups in pre_thread->skel_dump
444 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
445 walk_thread($rootset, $ctx, \&pre_thread);
448 return stream_thread($rootset, $ctx) unless $ctx->{flat};
450 # flat display: lazy load the full message from smsg
451 $ctx->{msgs} = $msgs;
452 $ctx->{-html_tip} = '<pre>';
453 PublicInbox::WwwStream::aresponse($ctx, 200, \&thread_html_i);
456 sub thread_html_i { # PublicInbox::WwwStream::getline callback
457 my ($ctx, $eml) = @_;
459 my $smsg = $ctx->{smsg};
460 if (exists $ctx->{-html_tip}) {
461 $ctx->{-title_html} = ascii_html($smsg->{subject});
462 $ctx->zmore($ctx->html_top);
464 return eml_entry($ctx, $eml);
466 while (my $smsg = shift @{$ctx->{msgs}}) {
467 return $smsg if exists($smsg->{blob});
469 my $skel = delete($ctx->{skel}) or return; # all done
475 sub multipart_text_as_html {
476 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
478 # scan through all parts, looking for displayable text
479 $_[0]->each_part(\&add_text_body, $_[1], 1);
482 sub submsg_hdr ($$) {
483 my ($ctx, $eml) = @_;
484 my $obfs_ibx = $ctx->{-obfs_ibx};
485 my $rv = $ctx->{obuf};
487 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
488 my @v = $eml->header($h);
490 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
497 sub attach_link ($$$$;$) {
498 my ($ctx, $ct, $p, $fn, $err) = @_;
499 my ($part, $depth, $idx) = @$p;
501 # Eml iteration clobbers multipart ->{bdy}, so do not offer
502 # downloads for 0-byte multipart attachments
503 return unless $part->{bdy};
505 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
506 my $size = bytes::length($part->body);
508 # hide attributes normally, unless we want to aid users in
509 # spotting MUA problems:
510 $ct =~ s/;.*// unless $err;
511 $ct = ascii_html($ct);
513 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
515 } elsif ($ct eq 'text/plain') {
520 my $rv = $ctx->{obuf};
521 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
524 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
527 $$rv .= "[-- Attachment #$idx: ";
528 my $ts = "Type: $ct, Size: $size bytes";
529 my $desc = $part->header('Content-Description') // $fn // '';
530 $desc = ascii_html($desc);
531 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
534 submsg_hdr($ctx, $part) if $part->{is_submsg};
539 sub add_text_body { # callback for each_part
541 my $upfx = $ctx->{mhref};
542 my $ibx = $ctx->{-inbox};
543 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
544 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
545 my ($part, $depth, $idx) = @$p;
546 my $ct = $part->content_type || 'text/plain';
547 my $fn = $part->filename;
548 my ($s, $err) = msg_part_text($part, $ct);
549 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
551 my $rv = $ctx->{obuf};
552 if ($part->{is_submsg}) {
553 submsg_hdr($ctx, $part);
557 # makes no difference to browsers, and don't screw up filename
558 # link generation in diffs with the extra '%0D'
561 # will be escaped to `•' in HTML
562 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
564 # always support diff-highlighting, but we can't linkify hunk
565 # headers for solver unless some coderepo are configured:
567 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
568 # diffstat anchors do not link across attachments or messages,
569 # -apfx is just a stable prefix for making diffstat anchors
570 # linkable to the first diff hunk w/o crossing attachments
571 $idx =~ tr!.!/!; # compatibility with previous versions
572 $ctx->{-apfx} = $upfx . $idx;
574 # do attr => filename mappings for diffstats in git diffs:
575 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
577 delete $ctx->{-long_path};
579 if ($ibx->{-repo_objs}) {
580 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
582 $spfx =~ s!/([^/]*)/\z!/!;
584 my $n_slash = $upfx =~ tr!/!/!;
587 } elsif ($n_slash == 1) {
589 } else { # nslash == 2
594 $ctx->{-spfx} = $spfx;
597 # some editors don't put trailing newlines at the end:
598 $s .= "\n" unless $s =~ /\n\z/s;
600 # split off quoted and unquoted blocks:
601 my @sections = PublicInbox::MsgIter::split_quotes($s);
602 undef $s; # free memory
603 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
604 # badly-encoded message with $err? tell the world about it!
605 attach_link($ctx, $ct, $p, $fn, $err);
608 foreach my $cur (@sections) {
610 # we use a <span> here to allow users to specify
611 # their own color for quoted text
612 $$rv .= qq(<span\nclass="q">);
613 $$rv .= $l->to_html($cur);
616 flush_diff($ctx, \$cur);
619 $$rv .= $l->to_html($cur);
621 undef $cur; # free memory
625 sub _msg_page_prepare_obuf {
626 my ($eml, $ctx) = @_;
627 my $over = $ctx->{-inbox}->over;
628 my $obfs_ibx = $ctx->{-obfs_ibx};
630 my $mids = mids_for_index($eml);
631 my $nr = $ctx->{nr}++;
632 if ($nr) { # unlikely
635 $ctx->{first_hdr} = $eml->header_obj;
638 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
640 $rv .= "<pre\nid=b>"; # anchor for body start
642 $ctx->{-upfx} = '../' if $over;
643 my @title; # (Subject[0], From[0])
644 for my $v ($eml->header('From')) {
645 my @n = PublicInbox::Address::names($v);
647 $title[1] //= ascii_html(join(', ', @n));
649 obfuscate_addrs($obfs_ibx, $v);
650 obfuscate_addrs($obfs_ibx, $title[1]);
652 $rv .= "From: $v\n" if $v ne '';
654 foreach my $h (qw(To Cc)) {
655 for my $v ($eml->header($h)) {
658 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
659 $rv .= "$h: $v\n" if $v ne '';
662 my @subj = $eml->header('Subject');
664 my $v = ascii_html(shift @subj);
665 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
667 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
669 for $v (@subj) { # multi-Subject message :<
671 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
672 $rv .= "Subject: $v\n";
674 } else { # dummy anchor for thread skeleton at bottom of page
675 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
676 $title[0] = '(no subject)';
678 for my $v ($eml->header('Date')) {
680 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
683 if (!$nr) { # first (and only) message, common case
684 $ctx->{-title_html} = join(' - ', @title);
685 $rv = $ctx->html_top . $rv;
687 if (scalar(@$mids) == 1) { # common case
688 my $mhtml = ascii_html($mids->[0]);
689 $rv .= "Message-ID: <$mhtml> ";
690 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
692 # X-Alt-Message-ID can happen if a message is injected from
693 # public-inbox-nntpd because of multiple Message-ID headers.
694 my $lnk = PublicInbox::Linkify->new;
696 for my $h (qw(Message-ID X-Alt-Message-ID)) {
697 $s .= "$h: $_\n" for ($eml->header_raw($h));
699 $lnk->linkify_mids('..', \$s, 1);
702 $rv .= _parent_headers($eml, $over);
708 qq(expand[<a\nhref="T/#u">flat</a>) .
709 qq(|<a\nhref="t/#u">nested</a>] ) .
710 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
711 qq(<a\nhref="t.atom">Atom feed</a>);
714 sub thread_skel ($$$) {
715 my ($skel, $ctx, $hdr) = @_;
716 my $mid = mids($hdr)->[0];
717 my $ibx = $ctx->{-inbox};
718 my ($nr, $msgs) = $ibx->over->get_thread($mid);
719 my $parent = in_reply_to($hdr);
720 $$skel .= "\n<b>Thread overview: </b>";
722 if (defined $parent) {
723 $$skel .= SKEL_EXPAND."\n ";
724 $$skel .= ghost_parent('../', $parent) . "\n";
726 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
728 $ctx->{next_msg} = undef;
729 $ctx->{parent_msg} = $parent;
734 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
736 # nb: mutt only shows the first Subject in the index pane
737 # when multiple Subject: headers are present, so we follow suit:
738 my $subj = $hdr->header('Subject') // '';
739 $subj = '(no subject)' if $subj eq '';
740 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
742 $ctx->{prev_attr} = '';
743 $ctx->{prev_level} = 0;
744 $ctx->{skel} = $skel;
746 # reduce hash lookups in skel_dump
747 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
748 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
750 $ctx->{parent_msg} = $parent;
753 sub _parent_headers {
754 my ($hdr, $over) = @_;
756 my @irt = $hdr->header_raw('In-Reply-To');
759 my $lnk = PublicInbox::Linkify->new;
760 $rv .= "In-Reply-To: $_\n" for @irt;
761 $lnk->linkify_mids('..', \$rv);
763 $refs = references($hdr);
764 my $irt = pop @$refs;
766 my $html = ascii_html($irt);
767 my $href = mid_href($irt);
768 $rv .= "In-Reply-To: <";
769 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
773 # do not display References: if search is present,
774 # we show the thread skeleton at the bottom, instead.
777 $refs //= references($hdr);
779 @$refs = map { linkify_ref_no_over($_) } @$refs;
780 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
785 # returns a string buffer via ->getline
787 my ($ctx, $hdr) = @_;
788 my $ibx = $ctx->{-inbox};
790 my $skel = " <a\nhref=\"$upfx\">index</a>";
794 thread_skel(\$skel, $ctx, $hdr);
799 if (my $n = $ctx->{next_msg}) {
801 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
804 my $par = $ctx->{parent_msg};
809 if (my $p = $ctx->{prev_msg}) {
810 $prev = mid_href($p);
811 if ($p && $par && $p eq $par) {
812 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
813 'rel=prev>prev parent</a>';
816 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
818 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
820 } elsif ($u) { # unlikely
821 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
823 $rv .= "$next $prev$parent ";
825 $rv .= qq(<a\nhref="#R">reply</a>);
828 $rv .= msg_reply($ctx, $hdr);
831 sub linkify_ref_no_over {
833 my $href = mid_href($mid);
834 my $html = ascii_html($mid);
835 "<<a\nhref=\"../$href/\">$html</a>>";
840 'm' . id_compress($msgid, 1);
844 my ($upfx, $mid) = @_;
846 my $href = mid_href($mid);
847 my $html = ascii_html($mid);
848 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
853 $level ? INDENT x ($level - 1) : '';
857 my ($ctx, $level, $node, $idx) = @_;
858 ++$ctx->{root_idx} if $level == 0;
859 if ($node->{mid} eq $ctx->{mid}) {
860 $ctx->{found_mid_at} = $ctx->{root_idx};
866 sub strict_loose_note ($) {
869 " -- strict thread matches above, loose matches on Subject: below --\n";
871 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
873 " -- use mbox.gz link to download all $nr messages --\n";
879 my ($ctx, $msgs) = @_;
880 require PublicInbox::SearchThread;
881 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
883 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
884 # and preserved in the future. This bug is hidden by `sid' matches
885 # in get_thread, so we never noticed it until now. And even when
886 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
887 # bump since reindexing is expensive and users may not do it
889 # loose threading could've returned too many results,
890 # put the root the message we care about at the top:
891 my $mid = $ctx->{mid};
892 if (defined($mid) && scalar(@$rootset) > 1) {
893 $ctx->{root_idx} = -1;
894 my $nr = scalar @$msgs;
895 walk_thread($rootset, $ctx, \&find_mid_root);
896 my $idx = $ctx->{found_mid_at};
897 if (defined($idx) && $idx != 0) {
898 my $tip = splice(@$rootset, $idx, 1);
899 @$rootset = reverse @$rootset;
900 unshift @$rootset, $tip;
901 $ctx->{sl_note} = strict_loose_note($nr);
909 require PublicInbox::ExtMsg;
910 PublicInbox::ExtMsg::ext_msg($ctx);
914 my ($prev_subj, $subj, $val) = @_;
916 my $omit; # '"' denotes identical text omitted
917 my (@prev_pop, @curr_pop);
918 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
919 push(@prev_pop, pop(@$prev_subj));
920 push(@curr_pop, pop(@$subj));
923 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
924 if (scalar(@curr_pop) == 1) {
926 push @$prev_subj, @prev_pop;
927 push @$subj, @curr_pop;
932 sub skel_dump { # walk_thread callback
933 my ($ctx, $level, $smsg) = @_;
934 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
936 my $skel = $ctx->{skel};
937 my $cur = $ctx->{cur};
938 my $mid = $smsg->{mid};
940 if ($level == 0 && $ctx->{skel_dump_roots}++) {
941 $$skel .= delete($ctx->{sl_note}) || '';
944 my $f = ascii_html($smsg->{from_name});
945 my $obfs_ibx = $ctx->{-obfs_ibx};
946 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
948 my $d = fmt_ts($smsg->{ds});
949 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
950 if (exists $ctx->{searchview}) {
951 if (defined(my $pct = $smsg->{pct})) {
952 $d .= (sprintf(' % 2u', $pct) . '%');
958 $d .= ' ' . indent_for($level) . th_pfx($level);
960 $ctx->{first_level} ||= $level;
962 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
963 $ctx->{prev_attr} = $attr;
965 $ctx->{prev_level} = $level;
970 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
971 "$attr [this message]</a></b>\n";
974 $ctx->{prev_msg} = $mid;
977 $ctx->{next_msg} ||= $mid;
980 # Subject is never undef, this mail was loaded from
981 # our Xapian which would've resulted in '' if it were
982 # really missing (and Filter rejects empty subjects)
983 my @subj = split(/ /, subject_normalized($smsg->{subject}));
984 # remove common suffixes from the subject if it matches the previous,
985 # so we do not show redundant text at the end.
986 my $prev_subj = $ctx->{prev_subj} || [];
987 $ctx->{prev_subj} = [ @subj ];
988 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
991 my $subj = join(' ', @subj);
992 $subj = ascii_html($subj);
993 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
994 $end = "$subj</a> $omit$f\n"
1000 my $mapping = $unmatched ? undef : $ctx->{mapping};
1002 my $map = $mapping->{$mid};
1003 $id = id_compress($mid, 1);
1005 $map->[0] = "$d<a\nhref=\"$m\">$end";
1008 $m = $ctx->{-upfx}.mid_href($mid).'/';
1010 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1015 my ($ctx, $level, $node) = @_;
1017 my $mid = $node->{mid};
1018 my $d = ' [not found] ';
1019 $d .= ' ' if exists $ctx->{searchview};
1020 $d .= indent_for($level) . th_pfx($level);
1021 my $upfx = $ctx->{-upfx};
1022 my $href = $upfx . mid_href($mid) . '/';
1023 my $html = ascii_html($mid);
1025 my $mapping = $ctx->{mapping};
1026 my $map = $mapping->{$mid} if $mapping;
1028 my $id = id_compress($mid, 1);
1029 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1030 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1032 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1034 ${$ctx->{skel}} .= $d;
1040 (eval { $a->topmost->{ds} } || 0) <=>
1041 (eval { $b->topmost->{ds} } || 0)
1045 # accumulate recent topics if search is supported
1046 # returns 200 if done, 404 if not
1047 sub acc_topic { # walk_thread callback
1048 my ($ctx, $level, $smsg) = @_;
1049 my $mid = $smsg->{mid};
1050 my $has_blob = $smsg->{blob} // do {
1051 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1052 %$smsg = (%$smsg, %$by_mid);
1057 my $subj = subject_normalized($smsg->{subject});
1058 $subj = '(no subject)' if $subj eq '';
1059 my $ds = $smsg->{ds};
1060 if ($level == 0) { # new, top-level topic
1061 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1062 $ctx->{-cur_topic} = $topic;
1063 push @{$ctx->{order}}, $topic;
1067 # continue existing topic
1068 my $topic = $ctx->{-cur_topic}; # should never be undef
1069 $topic->[0] = $ds if $ds > $topic->[0];
1070 $topic->[1]++; # bump N+ message counter
1071 my $seen = $topic->[2];
1072 if (scalar(@$topic) == 3) { # parent was a ghost
1073 push @$topic, $subj;
1074 } elsif (!defined($seen->{$subj})) {
1075 push @$topic, $level, $subj; # @extra messages
1077 $seen->{$subj} = $mid; # latest for subject
1078 } else { # ghost message
1079 return 1 if $level != 0; # ignore child ghosts
1080 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1081 push @{$ctx->{order}}, $topic;
1088 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1090 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1095 my $ibx = $ctx->{-inbox};
1096 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1098 # sort by recency, this allows new posts to "bump" old topics...
1099 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1100 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1102 next unless defined $top_subj; # ghost topic
1103 my $mid = delete $seen->{$top_subj};
1104 my $href = mid_href($mid);
1105 my $prev_subj = [ split(/ /, $top_subj) ];
1106 $top_subj = ascii_html($top_subj);
1109 # $n isn't the total number of posts on the topic,
1110 # just the number of posts in the current results window
1114 $anchor = '#u'; # top of only message
1116 $n = " ($n+ messages)";
1117 $anchor = '#t'; # thread skeleton
1120 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1121 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1122 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1123 " $ds UTC $n - $mbox / $atom\n";
1124 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1125 my $level = $extra[$i];
1126 my $subj = $extra[$i + 1]; # already normalized
1127 $mid = delete $seen->{$subj};
1128 my @subj = split(/ /, $subj);
1129 my @next_prev = @subj; # full copy
1130 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1131 $prev_subj = \@next_prev;
1132 $subj = join(' ', @subj);
1133 $subj = ascii_html($subj);
1134 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1135 $href = mid_href($mid);
1136 $s .= indent_for($level) . TCHILD;
1137 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1141 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1145 # only for the t= query parameter passed to overview DB
1146 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1149 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1150 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1153 sub pagination_footer ($$) {
1154 my ($ctx, $latest) = @_;
1155 delete $ctx->{qp} or return;
1156 my $next = $ctx->{next_page} || '';
1157 my $prev = $ctx->{prev_page} || '';
1159 $next = $next ? "$next " : ' ';
1160 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1162 "<hr><pre>page: $next$prev</pre>";
1165 sub paginate_recent ($$) {
1166 my ($ctx, $lim) = @_;
1167 my $t = $ctx->{qp}->{t} || '';
1168 my $opts = { limit => $lim };
1169 my ($after, $before);
1171 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1172 # if only $after exists "YYYYMMDD.." because "." could be skipped
1173 # if interpreted as an end-of-sentence
1174 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1175 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1177 my $ibx = $ctx->{-inbox};
1178 my $msgs = $ibx->recent($opts, $after, $before);
1179 my $nr = scalar @$msgs;
1180 if ($nr < $lim && defined($after)) {
1181 $after = $before = undef;
1182 $msgs = $ibx->recent($opts);
1183 $nr = scalar @$msgs;
1185 my $more = $nr == $lim;
1186 my ($newest, $oldest);
1188 $newest = $msgs->[0]->{ts};
1189 $oldest = $msgs->[-1]->{ts};
1190 # if we only had $after, our SQL query in ->recent ordered
1191 if ($newest < $oldest) {
1192 ($oldest, $newest) = ($newest, $oldest);
1193 $more = 0 if defined($after) && $after < $oldest;
1196 if (defined($oldest) && $more) {
1197 my $s = ts2str($oldest);
1198 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1200 if (defined($newest) && (defined($before) || defined($after))) {
1201 my $s = ts2str($newest);
1202 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1209 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1211 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1213 html_oneshot($ctx, dump_topics($ctx), \pagination_footer($ctx, '.'));
1217 sub thread_adj_level {
1218 my ($ctx, $level) = @_;
1220 my $max = $ctx->{cur_level};
1222 return ('', '') if $max == 0; # flat output
1224 # reset existing lists
1225 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1226 $ctx->{cur_level} = 0;
1228 } elsif ($level == $max) { # continue existing list
1230 } elsif ($level < $max) {
1231 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1232 $ctx->{cur_level} = $level;
1233 ("$beg<li>", '</li>');
1234 } else { # ($level > $max) # start a new level
1235 $ctx->{cur_level} = $level;
1236 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1241 sub ghost_index_entry {
1242 my ($ctx, $level, $node) = @_;
1243 my ($beg, $end) = thread_adj_level($ctx, $level);
1244 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})