1 # Copyright (C) 2014-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Used for displaying the HTML web interface.
5 # See Documentation/design_www.txt for this.
6 package PublicInbox::View;
9 use bytes (); # only for bytes::length
10 use PublicInbox::MsgTime qw(msg_datestamp);
11 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl mid_href);
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw(id_compress mids mids_for_index references
15 use PublicInbox::MsgIter;
16 use PublicInbox::Address;
17 use PublicInbox::WwwStream;
18 use PublicInbox::Reply;
19 use PublicInbox::ViewDiff qw(flush_diff);
21 use POSIX qw(strftime);
22 use Time::Local qw(timegm);
23 use PublicInbox::Smsg qw(subject_normalized);
24 use constant COLS => 72;
25 use constant INDENT => ' ';
26 use constant TCHILD => '` ';
27 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
31 if ($eml) { # called by WwwStream::async_eml or getline
32 my $smsg = $ctx->{smsg};
33 $ctx->{smsg} = $ctx->{over}->next_by_mid(@{$ctx->{next_arg}});
34 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
35 "../${\mid_href($smsg->{mid})}/" : '';
36 my $hdr = $eml->header_obj;
37 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
38 multipart_text_as_html($eml, $ctx);
40 $$obuf .= '</pre><hr>';
41 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
43 } else { # called by WwwStream::async_next or getline
44 $ctx->{smsg}; # may be undef
48 # /$INBOX/$MESSAGE_ID/ for unindexed v1 inboxes
51 my $eml = delete $ctx->{eml} or return;
52 my $hdr = $eml->header_obj;
54 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx);
55 multipart_text_as_html($eml, $ctx);
57 $$obuf .= '</pre><hr>';
58 eval { $$obuf .= html_footer($ctx, $hdr) };
62 sub no_over_html ($) {
64 my $bref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return; # 404
65 $ctx->{eml} = PublicInbox::Eml->new($bref);
66 PublicInbox::WwwStream::response($ctx, 200, \&no_over_i);
69 # public functions: (unstable)
73 my $ibx = $ctx->{-inbox};
74 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
75 my $over = $ctx->{over} = $ibx->over or return no_over_html($ctx);
77 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
78 $ctx->{smsg} = $over->next_by_mid(@$next_arg) or return; # undef == 404
79 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
82 # /$INBOX/$MESSAGE_ID/#R
86 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
88 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
91 my $ibx = $ctx->{-inbox};
92 if (my $url = $ibx->{infourl}) {
93 $url = prurl($ctx->{env}, $url);
94 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
97 my ($arg, $link, $reply_to_all) =
98 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
99 if (ref($arg) eq 'SCALAR') {
100 return '<pre id=R>'.ascii_html($$arg).'</pre>';
103 # mailto: link only works if address obfuscation is disabled
107 * If your mail client supports setting the <b>In-Reply-To</b> header
108 via mailto: links, try the <a
109 href="$link">mailto: link</a>
113 push @$arg, '/path/to/YOUR_REPLY';
114 $arg = ascii_html(join(" \\\n ", '', @$arg));
117 id=R><b>Reply instructions:</b>
119 You may reply publicly to <a
120 href=#t>this message</a> via plain-text email
121 using any one of the following methods:
123 * Save the following mbox file, import it into your mail client,
124 and $reply_to_all from there: <a
127 Avoid top-posting and favor interleaved quoting:
129 href="$p_url">$p_url</a>
131 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
132 switches of git-send-email(1):
137 href="$se_url">$se_url</a>
144 my $refs = references($hdr);
148 sub fold_addresses ($) {
149 return $_[0] if length($_[0]) <= COLS;
150 # try to fold on commas after non-word chars before $lim chars,
151 # Try to get the "," preceded by ">" or ")", but avoid folding
152 # on the comma where somebody uses "Lastname, Firstname".
153 # We also try to keep the last and penultimate addresses in
154 # the list on the same line if possible, hence the extra \z
155 # Fall back to folding on spaces at $lim + 1 chars
156 my $lim = COLS - 8; # 8 = "\t" display width
157 my $too_long = $lim + 1;
158 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
160 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
163 .{$too_long,}?)(?:\s|\z)/xgo));
166 sub _hdr_names_html ($$) {
167 my ($hdr, $field) = @_;
168 my @vals = $hdr->header($field) or return '';
169 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
173 my ($nr, $singular, $plural) = @_;
174 return "0 $plural" if $nr == 0;
175 $nr == 1 ? "$nr $singular" : "$nr $plural";
178 # human-friendly format
179 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
181 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
182 # this is already inside a <pre>
184 my ($ctx, $smsg, $eml, $more) = @_;
185 my $subj = delete $smsg->{subject};
186 my $mid_raw = $smsg->{mid};
187 my $id = id_compress($mid_raw, 1);
189 my $root_anchor = $ctx->{root_anchor} || '';
191 my $obfs_ibx = $ctx->{-obfs_ibx};
193 $subj = '(no subject)' if $subj eq '';
194 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
195 $subj = '<b>'.ascii_html($subj).'</b>';
196 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
197 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
199 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
201 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
203 # Deleting these fields saves about 400K as we iterate across 1K msgs
204 delete @$smsg{qw(ts blob)};
206 my $hdr = $eml->header_obj;
207 my $from = _hdr_names_html($hdr, 'From');
208 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
209 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
210 my $upfx = $ctx->{-upfx};
211 my $mhref = $upfx . mid_href($mid_raw) . '/';
212 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
213 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
214 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
215 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
216 my ($tlen, $clen) = (length($to), length($cc));
218 if (($tlen + $clen) > COLS) {
219 $to_cc .= ' To: '.$to."\n" if $tlen;
220 $to_cc .= ' Cc: '.$cc."\n" if $clen;
223 $to_cc .= ' To: '.$to;
224 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
226 $to_cc .= ' Cc: '.$cc if $clen;
230 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
233 my $mapping = $ctx->{mapping};
234 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
235 my $href = $upfx . mid_href($irt) . '/';
236 my $html = ascii_html($irt);
237 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
241 # scan through all parts, looking for displayable text
242 $ctx->{mhref} = $mhref;
244 $eml->each_part(\&add_text_body, $ctx, 1);
248 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
249 "<a\nhref=\"$mhref\">permalink</a>" .
250 " <a\nhref=\"${mhref}raw\">raw</a>" .
251 " <a\nhref=\"${mhref}#R\">reply</a>";
254 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
255 $rv .= "\t[relevance $pct%]";
258 my $nested = 'nested';
263 $flat = "<b>$flat</b>";
265 $nested = "<b>$nested</b>";
267 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
268 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
269 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
274 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
278 sub pad_link ($$;$) {
279 my ($mid, $level, $s) = @_;
281 my $id = id_compress($mid, 1);
282 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
286 my ($mid_raw, $irt, $id, $ctx) = @_;
288 my $mapping = $ctx->{mapping} or return $rv;
290 my $mid_map = $mapping->{$mid_raw};
292 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
293 my ($attr, $node, $idx, $level) = @$mid_map;
294 my $children = $node->{children};
295 my $nr_c = scalar @$children;
298 # delete saves about 200KB on a 1K message thread
299 if (my $refs = delete $node->{references}) {
300 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
302 my $irt_map = $mapping->{$$irt} if defined $$irt;
303 if (defined $irt_map) {
304 $siblings = $irt_map->[1]->{children};
305 $nr_s = scalar(@$siblings) - 1;
306 $rv .= $pad . $irt_map->[0];
308 my $prev = $siblings->[$idx - 1];
309 my $pmid = $prev->{mid};
311 my $s = ($idx - 1). ' preceding siblings ...';
312 $rv .= pad_link($pmid, $level, $s);
313 } elsif ($idx == 2) {
314 my $ppmid = $siblings->[0]->{mid};
315 $rv .= $pad . $mapping->{$ppmid}->[0];
317 $rv .= $pad . $mapping->{$pmid}->[0];
320 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
321 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
322 $attr =~ s!\n\z!</b>\n!s;
323 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
324 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
327 my $cmid = $children->[0]->{mid};
328 $rv .= $pad . $mapping->{$cmid}->[0];
330 my $s = ($nr_c - 1). ' more replies';
331 $rv .= pad_link($cmid, $level + 1, $s);
332 } elsif (my $cn = $children->[1]) {
333 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
337 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
339 my $nmid = $next->{mid};
340 $rv .= $pad . $mapping->{$nmid}->[0];
341 my $nnext = $nr_s - $idx;
343 my $s = ($nnext - 1).' subsequent siblings';
344 $rv .= pad_link($nmid, $level, $s);
345 } elsif (my $nn = $siblings->[$idx + 2]) {
346 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
349 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
352 # non-recursive thread walker
353 sub walk_thread ($$$) {
354 my ($rootset, $ctx, $cb) = @_;
355 my @q = map { (0, $_, -1) } @$rootset;
357 my ($level, $node, $i) = splice(@q, 0, 3);
358 defined $node or next;
359 $cb->($ctx, $level, $node, $i) or return;
362 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
366 sub pre_thread { # walk_thread callback
367 my ($ctx, $level, $node, $idx) = @_;
368 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
369 skel_dump($ctx, $level, $node);
372 sub thread_eml_entry {
373 my ($ctx, $level, $smsg, $eml) = @_;
374 my ($beg, $end) = thread_adj_level($ctx, $level);
375 $beg . '<pre>' . eml_entry($ctx, $smsg, $eml, 0) . '</pre>' . $end;
378 sub next_in_queue ($;$) {
379 my ($q, $ghost_ok) = @_;
381 my ($level, $smsg) = splice(@$q, 0, 2);
383 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
384 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
389 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
391 return unless exists($ctx->{skel});
392 my $nr = $ctx->{nr}++;
393 my ($level, $smsg) = next_in_queue($ctx->{-queue}, $nr);
396 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{skel}};
398 my $eml = $ctx->{-inbox}->smsg_eml($smsg) or return
399 ghost_index_entry($ctx, $level, $smsg);
402 $ctx->{-title_html} = ascii_html($smsg->{subject});
403 $ctx->html_top . thread_eml_entry($ctx, $level, $smsg, $eml);
405 thread_eml_entry($ctx, $level, $smsg, $eml);
409 sub stream_thread ($$) {
410 my ($rootset, $ctx) = @_;
411 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
412 PublicInbox::WwwStream::response($ctx, 200, \&stream_thread_i);
415 # /$INBOX/$MESSAGE_ID/t/
418 my $mid = $ctx->{mid};
419 my $ibx = $ctx->{-inbox};
420 my ($nr, $msgs) = $ibx->over->get_thread($mid);
421 return missing_thread($ctx) if $nr == 0;
422 my $skel = '<hr><pre>';
423 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
424 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
425 $skel .= "<b\nid=t>Thread overview:</b> ";
426 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
427 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
428 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
429 $skel .= "-- links below jump to the message on this page --\n";
430 $ctx->{-upfx} = '../../';
431 $ctx->{cur_level} = 0;
432 $ctx->{skel} = \$skel;
433 $ctx->{prev_attr} = '';
434 $ctx->{prev_level} = 0;
435 $ctx->{root_anchor} = anchor_for($mid);
436 $ctx->{mapping} = {};
437 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
440 my $rootset = thread_results($ctx, $msgs);
442 # reduce hash lookups in pre_thread->skel_dump
443 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
444 walk_thread($rootset, $ctx, \&pre_thread);
447 return stream_thread($rootset, $ctx) unless $ctx->{flat};
449 # flat display: lazy load the full message from smsg
450 $ctx->{msgs} = $msgs;
451 $ctx->{-html_tip} = '<pre>';
452 PublicInbox::WwwStream::response($ctx, 200, \&thread_html_i);
455 sub thread_html_i { # PublicInbox::WwwStream::getline callback
457 my $msgs = $ctx->{msgs} or return;
458 while (my $smsg = shift @$msgs) {
459 my $eml = $ctx->{-inbox}->smsg_eml($smsg) or next;
460 if (exists $ctx->{-html_tip}) {
461 $ctx->{-title_html} = ascii_html($smsg->{subject});
462 return $ctx->html_top .
463 eml_entry($ctx, $smsg, $eml, scalar @$msgs);
465 return eml_entry($ctx, $smsg, $eml, scalar @$msgs);
467 my ($skel) = delete @$ctx{qw(skel msgs)};
471 sub multipart_text_as_html {
472 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
474 # scan through all parts, looking for displayable text
475 $_[0]->each_part(\&add_text_body, $_[1], 1);
478 sub submsg_hdr ($$) {
479 my ($ctx, $eml) = @_;
480 my $obfs_ibx = $ctx->{-obfs_ibx};
481 my $rv = $ctx->{obuf};
483 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
484 my @v = $eml->header($h);
486 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
493 sub attach_link ($$$$;$) {
494 my ($ctx, $ct, $p, $fn, $err) = @_;
495 my ($part, $depth, $idx) = @$p;
497 # Eml iteration clobbers multipart ->{bdy}, so do not offer
498 # downloads for 0-byte multipart attachments
499 return unless $part->{bdy};
501 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
502 my $size = bytes::length($part->body);
504 # hide attributes normally, unless we want to aid users in
505 # spotting MUA problems:
506 $ct =~ s/;.*// unless $err;
507 $ct = ascii_html($ct);
509 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
511 } elsif ($ct eq 'text/plain') {
516 my $rv = $ctx->{obuf};
517 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
520 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
523 $$rv .= "[-- Attachment #$idx: ";
524 my $ts = "Type: $ct, Size: $size bytes";
525 my $desc = $part->header('Content-Description') // $fn // '';
526 $desc = ascii_html($desc);
527 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
530 submsg_hdr($ctx, $part) if $part->{is_submsg};
535 sub add_text_body { # callback for each_part
537 my $upfx = $ctx->{mhref};
538 my $ibx = $ctx->{-inbox};
539 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
540 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
541 my ($part, $depth, $idx) = @$p;
542 my $ct = $part->content_type || 'text/plain';
543 my $fn = $part->filename;
544 my ($s, $err) = msg_part_text($part, $ct);
545 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
547 my $rv = $ctx->{obuf};
548 if ($part->{is_submsg}) {
549 submsg_hdr($ctx, $part);
553 # makes no difference to browsers, and don't screw up filename
554 # link generation in diffs with the extra '%0D'
557 # will be escaped to `•' in HTML
558 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
560 # always support diff-highlighting, but we can't linkify hunk
561 # headers for solver unless some coderepo are configured:
563 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
564 # diffstat anchors do not link across attachments or messages,
565 # -apfx is just a stable prefix for making diffstat anchors
566 # linkable to the first diff hunk w/o crossing attachments
567 $idx =~ tr!.!/!; # compatibility with previous versions
568 $ctx->{-apfx} = $upfx . $idx;
570 # do attr => filename mappings for diffstats in git diffs:
571 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
573 delete $ctx->{-long_path};
575 if ($ibx->{-repo_objs}) {
576 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
578 $spfx =~ s!/([^/]*)/\z!/!;
580 my $n_slash = $upfx =~ tr!/!/!;
583 } elsif ($n_slash == 1) {
585 } else { # nslash == 2
590 $ctx->{-spfx} = $spfx;
593 # some editors don't put trailing newlines at the end:
594 $s .= "\n" unless $s =~ /\n\z/s;
596 # split off quoted and unquoted blocks:
597 my @sections = PublicInbox::MsgIter::split_quotes($s);
598 undef $s; # free memory
599 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
600 # badly-encoded message with $err? tell the world about it!
601 attach_link($ctx, $ct, $p, $fn, $err);
604 foreach my $cur (@sections) {
606 # we use a <span> here to allow users to specify
607 # their own color for quoted text
608 $$rv .= qq(<span\nclass="q">);
609 $$rv .= $l->to_html($cur);
612 flush_diff($ctx, \$cur);
615 $$rv .= $l->to_html($cur);
617 undef $cur; # free memory
621 sub _msg_page_prepare_obuf {
622 my ($hdr, $ctx) = @_;
623 my $over = $ctx->{-inbox}->over;
624 my $obfs_ibx = $ctx->{-obfs_ibx};
626 my $mids = mids_for_index($hdr);
627 my $nr = $ctx->{nr}++;
628 if ($nr) { # unlikely
631 $ctx->{first_hdr} = $hdr;
634 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
636 $rv .= "<pre\nid=b>"; # anchor for body start
638 $ctx->{-upfx} = '../' if $over;
639 my @title; # (Subject[0], From[0])
640 for my $v ($hdr->header('From')) {
641 my @n = PublicInbox::Address::names($v);
643 $title[1] //= ascii_html(join(', ', @n));
645 obfuscate_addrs($obfs_ibx, $v);
646 obfuscate_addrs($obfs_ibx, $title[1]);
648 $rv .= "From: $v\n" if $v ne '';
650 foreach my $h (qw(To Cc)) {
651 for my $v ($hdr->header($h)) {
654 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
655 $rv .= "$h: $v\n" if $v ne '';
658 my @subj = $hdr->header('Subject');
660 my $v = ascii_html(shift @subj);
661 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
663 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
665 for $v (@subj) { # multi-Subject message :<
667 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
668 $rv .= "Subject: $v\n";
670 } else { # dummy anchor for thread skeleton at bottom of page
671 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
672 $title[0] = '(no subject)';
674 for my $v ($hdr->header('Date')) {
676 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
679 if (!$nr) { # first (and only) message, common case
680 $ctx->{-title_html} = join(' - ', @title);
681 $rv = $ctx->html_top . $rv;
683 if (scalar(@$mids) == 1) { # common case
684 my $mhtml = ascii_html($mids->[0]);
685 $rv .= "Message-ID: <$mhtml> ";
686 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
688 # X-Alt-Message-ID can happen if a message is injected from
689 # public-inbox-nntpd because of multiple Message-ID headers.
690 my $lnk = PublicInbox::Linkify->new;
692 for my $h (qw(Message-ID X-Alt-Message-ID)) {
693 $s .= "$h: $_\n" for ($hdr->header_raw($h));
695 $lnk->linkify_mids('..', \$s, 1);
698 $rv .= _parent_headers($hdr, $over);
704 qq(expand[<a\nhref="T/#u">flat</a>) .
705 qq(|<a\nhref="t/#u">nested</a>] ) .
706 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
707 qq(<a\nhref="t.atom">Atom feed</a>);
710 sub thread_skel ($$$) {
711 my ($skel, $ctx, $hdr) = @_;
712 my $mid = mids($hdr)->[0];
713 my $ibx = $ctx->{-inbox};
714 my ($nr, $msgs) = $ibx->over->get_thread($mid);
715 my $parent = in_reply_to($hdr);
716 $$skel .= "\n<b>Thread overview: </b>";
718 if (defined $parent) {
719 $$skel .= SKEL_EXPAND."\n ";
720 $$skel .= ghost_parent('../', $parent) . "\n";
722 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
724 $ctx->{next_msg} = undef;
725 $ctx->{parent_msg} = $parent;
730 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
732 # nb: mutt only shows the first Subject in the index pane
733 # when multiple Subject: headers are present, so we follow suit:
734 my $subj = $hdr->header('Subject') // '';
735 $subj = '(no subject)' if $subj eq '';
736 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
738 $ctx->{prev_attr} = '';
739 $ctx->{prev_level} = 0;
740 $ctx->{skel} = $skel;
742 # reduce hash lookups in skel_dump
743 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
744 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
746 $ctx->{parent_msg} = $parent;
749 sub _parent_headers {
750 my ($hdr, $over) = @_;
752 my @irt = $hdr->header_raw('In-Reply-To');
755 my $lnk = PublicInbox::Linkify->new;
756 $rv .= "In-Reply-To: $_\n" for @irt;
757 $lnk->linkify_mids('..', \$rv);
759 $refs = references($hdr);
760 my $irt = pop @$refs;
762 my $html = ascii_html($irt);
763 my $href = mid_href($irt);
764 $rv .= "In-Reply-To: <";
765 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
769 # do not display References: if search is present,
770 # we show the thread skeleton at the bottom, instead.
773 $refs //= references($hdr);
775 @$refs = map { linkify_ref_no_over($_) } @$refs;
776 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
781 # returns a string buffer via ->getline
783 my ($ctx, $hdr) = @_;
784 my $ibx = $ctx->{-inbox};
786 my $skel = " <a\nhref=\"$upfx\">index</a>";
790 thread_skel(\$skel, $ctx, $hdr);
795 if (my $n = $ctx->{next_msg}) {
797 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
800 my $par = $ctx->{parent_msg};
805 if (my $p = $ctx->{prev_msg}) {
806 $prev = mid_href($p);
807 if ($p && $par && $p eq $par) {
808 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
809 'rel=prev>prev parent</a>';
812 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
814 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
816 } elsif ($u) { # unlikely
817 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
819 $rv .= "$next $prev$parent ";
821 $rv .= qq(<a\nhref="#R">reply</a>);
824 $rv .= msg_reply($ctx, $hdr);
827 sub linkify_ref_no_over {
829 my $href = mid_href($mid);
830 my $html = ascii_html($mid);
831 "<<a\nhref=\"../$href/\">$html</a>>";
836 'm' . id_compress($msgid, 1);
840 my ($upfx, $mid) = @_;
842 my $href = mid_href($mid);
843 my $html = ascii_html($mid);
844 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
849 $level ? INDENT x ($level - 1) : '';
853 my ($ctx, $level, $node, $idx) = @_;
854 ++$ctx->{root_idx} if $level == 0;
855 if ($node->{mid} eq $ctx->{mid}) {
856 $ctx->{found_mid_at} = $ctx->{root_idx};
862 sub strict_loose_note ($) {
865 " -- strict thread matches above, loose matches on Subject: below --\n";
867 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
869 " -- use mbox.gz link to download all $nr messages --\n";
875 my ($ctx, $msgs) = @_;
876 require PublicInbox::SearchThread;
877 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
879 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
880 # and preserved in the future. This bug is hidden by `sid' matches
881 # in get_thread, so we never noticed it until now. And even when
882 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
883 # bump since reindexing is expensive and users may not do it
885 # loose threading could've returned too many results,
886 # put the root the message we care about at the top:
887 my $mid = $ctx->{mid};
888 if (defined($mid) && scalar(@$rootset) > 1) {
889 $ctx->{root_idx} = -1;
890 my $nr = scalar @$msgs;
891 walk_thread($rootset, $ctx, \&find_mid_root);
892 my $idx = $ctx->{found_mid_at};
893 if (defined($idx) && $idx != 0) {
894 my $tip = splice(@$rootset, $idx, 1);
895 @$rootset = reverse @$rootset;
896 unshift @$rootset, $tip;
897 $ctx->{sl_note} = strict_loose_note($nr);
905 require PublicInbox::ExtMsg;
906 PublicInbox::ExtMsg::ext_msg($ctx);
910 my ($prev_subj, $subj, $val) = @_;
912 my $omit; # '"' denotes identical text omitted
913 my (@prev_pop, @curr_pop);
914 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
915 push(@prev_pop, pop(@$prev_subj));
916 push(@curr_pop, pop(@$subj));
919 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
920 if (scalar(@curr_pop) == 1) {
922 push @$prev_subj, @prev_pop;
923 push @$subj, @curr_pop;
928 sub skel_dump { # walk_thread callback
929 my ($ctx, $level, $smsg) = @_;
930 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
932 my $skel = $ctx->{skel};
933 my $cur = $ctx->{cur};
934 my $mid = $smsg->{mid};
936 if ($level == 0 && $ctx->{skel_dump_roots}++) {
937 $$skel .= delete($ctx->{sl_note}) || '';
940 my $f = ascii_html($smsg->{from_name});
941 my $obfs_ibx = $ctx->{-obfs_ibx};
942 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
944 my $d = fmt_ts($smsg->{ds});
945 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
946 if (exists $ctx->{searchview}) {
947 if (defined(my $pct = $smsg->{pct})) {
948 $d .= (sprintf(' % 2u', $pct) . '%');
954 $d .= ' ' . indent_for($level) . th_pfx($level);
956 $ctx->{first_level} ||= $level;
958 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
959 $ctx->{prev_attr} = $attr;
961 $ctx->{prev_level} = $level;
966 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
967 "$attr [this message]</a></b>\n";
970 $ctx->{prev_msg} = $mid;
973 $ctx->{next_msg} ||= $mid;
976 # Subject is never undef, this mail was loaded from
977 # our Xapian which would've resulted in '' if it were
978 # really missing (and Filter rejects empty subjects)
979 my @subj = split(/ /, subject_normalized($smsg->{subject}));
980 # remove common suffixes from the subject if it matches the previous,
981 # so we do not show redundant text at the end.
982 my $prev_subj = $ctx->{prev_subj} || [];
983 $ctx->{prev_subj} = [ @subj ];
984 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
987 my $subj = join(' ', @subj);
988 $subj = ascii_html($subj);
989 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
990 $end = "$subj</a> $omit$f\n"
996 my $mapping = $unmatched ? undef : $ctx->{mapping};
998 my $map = $mapping->{$mid};
999 $id = id_compress($mid, 1);
1001 $map->[0] = "$d<a\nhref=\"$m\">$end";
1004 $m = $ctx->{-upfx}.mid_href($mid).'/';
1006 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1011 my ($ctx, $level, $node) = @_;
1013 my $mid = $node->{mid};
1014 my $d = ' [not found] ';
1015 $d .= ' ' if exists $ctx->{searchview};
1016 $d .= indent_for($level) . th_pfx($level);
1017 my $upfx = $ctx->{-upfx};
1018 my $href = $upfx . mid_href($mid) . '/';
1019 my $html = ascii_html($mid);
1021 my $mapping = $ctx->{mapping};
1022 my $map = $mapping->{$mid} if $mapping;
1024 my $id = id_compress($mid, 1);
1025 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1026 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1028 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1030 ${$ctx->{skel}} .= $d;
1036 (eval { $a->topmost->{ds} } || 0) <=>
1037 (eval { $b->topmost->{ds} } || 0)
1041 # accumulate recent topics if search is supported
1042 # returns 200 if done, 404 if not
1043 sub acc_topic { # walk_thread callback
1044 my ($ctx, $level, $smsg) = @_;
1045 my $mid = $smsg->{mid};
1046 my $has_blob = $smsg->{blob} // do {
1047 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1048 %$smsg = (%$smsg, %$by_mid);
1053 my $subj = subject_normalized($smsg->{subject});
1054 $subj = '(no subject)' if $subj eq '';
1055 my $ds = $smsg->{ds};
1056 if ($level == 0) { # new, top-level topic
1057 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1058 $ctx->{-cur_topic} = $topic;
1059 push @{$ctx->{order}}, $topic;
1063 # continue existing topic
1064 my $topic = $ctx->{-cur_topic}; # should never be undef
1065 $topic->[0] = $ds if $ds > $topic->[0];
1066 $topic->[1]++; # bump N+ message counter
1067 my $seen = $topic->[2];
1068 if (scalar(@$topic) == 3) { # parent was a ghost
1069 push @$topic, $subj;
1070 } elsif (!defined($seen->{$subj})) {
1071 push @$topic, $level, $subj; # @extra messages
1073 $seen->{$subj} = $mid; # latest for subject
1074 } else { # ghost message
1075 return 1 if $level != 0; # ignore child ghosts
1076 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1077 push @{$ctx->{order}}, $topic;
1084 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1086 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1091 my $ibx = $ctx->{-inbox};
1092 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1094 # sort by recency, this allows new posts to "bump" old topics...
1095 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1096 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1098 next unless defined $top_subj; # ghost topic
1099 my $mid = delete $seen->{$top_subj};
1100 my $href = mid_href($mid);
1101 my $prev_subj = [ split(/ /, $top_subj) ];
1102 $top_subj = ascii_html($top_subj);
1105 # $n isn't the total number of posts on the topic,
1106 # just the number of posts in the current results window
1110 $anchor = '#u'; # top of only message
1112 $n = " ($n+ messages)";
1113 $anchor = '#t'; # thread skeleton
1116 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1117 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1118 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1119 " $ds UTC $n - $mbox / $atom\n";
1120 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1121 my $level = $extra[$i];
1122 my $subj = $extra[$i + 1]; # already normalized
1123 $mid = delete $seen->{$subj};
1124 my @subj = split(/ /, $subj);
1125 my @next_prev = @subj; # full copy
1126 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1127 $prev_subj = \@next_prev;
1128 $subj = join(' ', @subj);
1129 $subj = ascii_html($subj);
1130 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1131 $href = mid_href($mid);
1132 $s .= indent_for($level) . TCHILD;
1133 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1137 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1141 # only for the t= query parameter passed to overview DB
1142 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1145 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1146 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1149 sub pagination_footer ($$) {
1150 my ($ctx, $latest) = @_;
1151 delete $ctx->{qp} or return;
1152 my $next = $ctx->{next_page} || '';
1153 my $prev = $ctx->{prev_page} || '';
1155 $next = $next ? "$next " : ' ';
1156 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1158 "<hr><pre>page: $next$prev</pre>";
1161 sub index_nav { # callback for WwwStream::getline
1163 return $ctx->html_top if exists $ctx->{-html_tip};
1164 pagination_footer($ctx, '.')
1167 sub paginate_recent ($$) {
1168 my ($ctx, $lim) = @_;
1169 my $t = $ctx->{qp}->{t} || '';
1170 my $opts = { limit => $lim };
1171 my ($after, $before);
1173 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1174 # if only $after exists "YYYYMMDD.." because "." could be skipped
1175 # if interpreted as an end-of-sentence
1176 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1177 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1179 my $ibx = $ctx->{-inbox};
1180 my $msgs = $ibx->recent($opts, $after, $before);
1181 my $nr = scalar @$msgs;
1182 if ($nr < $lim && defined($after)) {
1183 $after = $before = undef;
1184 $msgs = $ibx->recent($opts);
1185 $nr = scalar @$msgs;
1187 my $more = $nr == $lim;
1188 my ($newest, $oldest);
1190 $newest = $msgs->[0]->{ts};
1191 $oldest = $msgs->[-1]->{ts};
1192 # if we only had $after, our SQL query in ->recent ordered
1193 if ($newest < $oldest) {
1194 ($oldest, $newest) = ($newest, $oldest);
1195 $more = 0 if defined($after) && $after < $oldest;
1198 if (defined($oldest) && $more) {
1199 my $s = ts2str($oldest);
1200 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1202 if (defined($newest) && (defined($before) || defined($after))) {
1203 my $s = ts2str($newest);
1204 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1211 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1213 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1215 PublicInbox::WwwStream::response($ctx, dump_topics($ctx), \&index_nav);
1218 sub thread_adj_level {
1219 my ($ctx, $level) = @_;
1221 my $max = $ctx->{cur_level};
1223 return ('', '') if $max == 0; # flat output
1225 # reset existing lists
1226 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1227 $ctx->{cur_level} = 0;
1229 } elsif ($level == $max) { # continue existing list
1231 } elsif ($level < $max) {
1232 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1233 $ctx->{cur_level} = $level;
1234 ("$beg<li>", '</li>');
1235 } else { # ($level > $max) # start a new level
1236 $ctx->{cur_level} = $level;
1237 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1242 sub ghost_index_entry {
1243 my ($ctx, $level, $node) = @_;
1244 my ($beg, $end) = thread_adj_level($ctx, $level);
1245 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})