1 # Copyright (C) 2014, Eric Wong <normalperson@yhbt.net> and all contributors
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
3 package PublicInbox::View;
6 use URI::Escape qw/uri_escape_utf8/;
7 use Date::Parse qw/str2time/;
8 use Encode qw/find_encoding/;
9 use Encode::MIME::Header;
10 use Email::MIME::ContentType qw/parse_content_type/;
11 use PublicInbox::Hval;
12 use PublicInbox::MID qw/mid_clean mid_compressed/;
16 # TODO: make these constants tunable
17 use constant MAX_INLINE_QUOTED => 12; # half an 80x24 terminal
18 use constant MAX_TRUNC_LEN => 72;
19 use constant PRE_WRAP => "<pre\nstyle=\"white-space:pre-wrap\">";
20 use constant T_ANCHOR => '#u';
22 *ascii_html = *PublicInbox::Hval::ascii_html;
24 my $enc_utf8 = find_encoding('UTF-8');
28 my ($class, $mime, $full_pfx, $footer, $srch) = @_;
29 if (defined $footer) {
30 $footer = "\n" . $footer;
34 headers_to_html_header($mime, $full_pfx, $srch) .
35 multipart_text_as_html($mime, $full_pfx) .
36 '</pre><hr />' . PRE_WRAP .
37 html_footer($mime, 1, $full_pfx, $srch) . $footer .
38 '</pre></body></html>';
42 my ($class, $mime, $full_pfx) = @_;
44 PRE_WRAP . multipart_text_as_html($mime, $full_pfx) . '</pre>';
47 # this is already inside a <pre>
48 # state = [ time, seen = {}, first_commit, page_nr = 0 ]
50 my (undef, $mime, $level, $state) = @_;
51 my (undef, $seen, $first_commit) = @$state;
52 my $midx = $state->[3]++;
53 my ($prev, $next) = ($midx - 1, $midx + 1);
56 my $enc_msg = enc_for($mime->header("Content-Type"));
57 my $subj = $mime->header('Subject');
58 my $header_obj = $mime->header_obj;
60 my $mid_raw = $header_obj->header_raw('Message-ID');
61 my $id = anchor_for($mid_raw);
62 $seen->{$id} = "#$id"; # save the anchor for later
64 my $mid = PublicInbox::Hval->new_msgid($mid_raw);
65 my $from = PublicInbox::Hval->new_oneline($mime->header('From'))->raw;
66 my @from = Email::Address->parse($from);
67 $from = $from[0]->name;
68 (defined($from) && length($from)) or $from = $from[0]->address;
70 $from = PublicInbox::Hval->new_oneline($from)->as_html;
71 $subj = PublicInbox::Hval->new_oneline($subj)->as_html;
72 my $pfx = (' ' x $level);
73 my $root_anchor = $seen->{root_anchor};
75 my $more = 'permalink';
78 $subj = "<u\nid=\"u\">$subj</u>" if $root_anchor eq $id;
83 my $ts = $mime->header('X-PI-TS');
84 my $fmt = '%Y-%m-%d %H:%M UTC';
85 $ts = POSIX::strftime($fmt, gmtime($ts));
87 $rv .= "$pfx<b\nid=\"$id\">$subj</b>\n$pfx";
88 $rv .= "- by $from @ $ts - ";
89 $rv .= "<a\nid=\"s$midx\"\nhref=\"#s$next\">next</a>";
91 $rv .= "/<a\nhref=\"#s$prev\">prev</a>";
95 my $irt = $header_obj->header_raw('In-Reply-To');
96 my ($anchor_idx, $anchor, $t_anchor);
98 $anchor_idx = anchor_for($irt);
99 $anchor = $seen->{$anchor_idx};
100 $t_anchor = T_ANCHOR;
104 my $href = $mid->as_href;
105 my $mhref = "${path}m/$href.html";
106 my $fhref = "${path}f/$href.html";
107 # scan through all parts, looking for displayable text
108 $mime->walk_parts(sub {
109 $rv .= index_walk($_[0], $pfx, $enc_msg, $part_nr, $fhref,
114 $rv .= "\n$pfx<a\nhref=\"$mhref\">$more</a> ";
115 my $txt = "${path}m/$href.txt";
116 $rv .= "<a\nhref=\"$txt\">raw</a> ";
117 $rv .= html_footer($mime, 0);
120 unless (defined $anchor) {
121 my $v = PublicInbox::Hval->new_msgid($irt);
123 $anchor = "${path}m/$v.html";
124 $seen->{$anchor_idx} = $anchor;
126 $rv .= " <a\nhref=\"$anchor\">parent</a>";
130 $rv .= " <a\nhref=\"t/$href.html$t_anchor\">thread</a>";
137 my (undef, $ctx, $foot, $srch) = @_;
138 my $mid = mid_compressed($ctx->{mid});
139 my $res = $srch->get_thread($mid);
141 require PublicInbox::GitCatFile;
142 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
143 my $nr = scalar @{$res->{msgs}};
144 return $rv if $nr == 0;
146 while (my $smsg = shift @{$res->{msgs}}) {
149 # Duplicated from WWW.pm
150 my ($x2, $x38) = ($m =~ /\A([a-f0-9]{2})([a-f0-9]{38})\z/);
152 unless (defined $x38) {
154 $m = Digest::SHA::sha1_hex($m);
155 ($x2, $x38) = ($m =~ /\A([a-f0-9]{2})([a-f0-9]{38})\z/);
158 # FIXME: duplicated code from Feed.pm
160 my $str = $git->cat_file("HEAD:$x2/$x38");
161 Email::MIME->new($str);
164 my $t = eval { str2time($mime->header('Date')) };
165 defined($t) or $t = 0;
166 $mime->header_set('X-PI-TS', $t);
170 require PublicInbox::Thread;
171 my $th = PublicInbox::Thread->new(@msgs);
173 $th->order(*PublicInbox::Thread::sort_ts);
174 my $state = [ undef, { root_anchor => anchor_for($mid) }, undef, 0 ];
175 thread_entry(\$rv, $state, $_, 0) for $th->rootset;
176 my $final_anchor = $state->[3];
177 my $next = "<a\nid=\"s$final_anchor\">end of thread</a>\n";
179 $rv .= "</pre><hr />" . PRE_WRAP . $next . $foot . "</pre>";
182 sub subject_path_html {
183 my (undef, $ctx, $foot, $srch) = @_;
184 my $path = $ctx->{subject_path};
185 my $res = $srch->get_subject_path($path);
187 require PublicInbox::GitCatFile;
188 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
189 my $nr = scalar @{$res->{msgs}};
190 return $rv if $nr == 0;
192 while (my $smsg = shift @{$res->{msgs}}) {
195 # Duplicated from WWW.pm
196 my ($x2, $x38) = ($m =~ /\A([a-f0-9]{2})([a-f0-9]{38})\z/);
198 unless (defined $x38) {
200 $m = Digest::SHA::sha1_hex($m);
201 ($x2, $x38) = ($m =~ /\A([a-f0-9]{2})([a-f0-9]{38})\z/);
204 # FIXME: duplicated code from Feed.pm
206 my $str = $git->cat_file("HEAD:$x2/$x38");
207 Email::MIME->new($str);
210 my $t = eval { str2time($mime->header('Date')) };
211 defined($t) or $t = 0;
212 $mime->header_set('X-PI-TS', $t);
216 require PublicInbox::Thread;
217 my $th = PublicInbox::Thread->new(@msgs);
219 $th->order(*PublicInbox::Thread::sort_ts);
220 my $state = [ undef, { root_anchor => 'dummy' }, undef, 0 ];
221 thread_entry(\$rv, $state, $_, 0) for $th->rootset;
222 my $final_anchor = $state->[3];
223 my $next = "<a\nid=\"s$final_anchor\">end of thread</a>\n";
225 $rv .= "</pre><hr />" . PRE_WRAP . $next . $foot . "</pre>";
228 # only private functions below.
231 my ($part, $pfx, $enc_msg, $part_nr, $fhref, $more) = @_;
233 return $rv if $part->subparts; # walk_parts already recurses
234 my $ct = $part->content_type;
236 # account for filter bugs...
237 return if defined $ct && $ct =~ m!\btext/[xh]+tml\b!i;
239 my $enc = enc_for($ct, $enc_msg);
242 my $fn = $part->filename;
243 defined($fn) or $fn = "part #" . ($part_nr + 1);
244 $rv .= $pfx . add_filename_line($enc->decode($fn));
247 my $s = add_text_body_short($enc, $part, $part_nr, $fhref);
249 # drop the remainder of git patches, they're usually better
250 # to review when the full message is viewed
251 $s =~ s!^---+\n.*\z!!ms and $$more = 'more...';
254 $s =~ s/^-- \n.*\z//ms and $$more = 'more...';
256 # kill any leading or trailing whitespace
270 my ($ct, $default) = @_;
271 $default ||= $enc_utf8;
272 defined $ct or return $default;
273 my $ct_parsed = parse_content_type($ct);
275 if (my $charset = $ct_parsed->{attributes}->{charset}) {
276 my $enc = find_encoding($charset);
283 sub multipart_text_as_html {
284 my ($mime, $full_pfx, $srch) = @_;
287 my $enc_msg = enc_for($mime->header("Content-Type"));
289 # scan through all parts, looking for displayable text
290 $mime->walk_parts(sub {
292 return if $part->subparts; # walk_parts already recurses
293 my $ct = $part->content_type;
295 # account for filter bugs...
296 return if defined $ct && $ct =~ m!\btext/[xh]+tml\b!i;
298 my $enc = enc_for($ct, $enc_msg);
301 my $fn = $part->filename;
302 defined($fn) or $fn = "part #" . ($part_nr + 1);
303 $rv .= add_filename_line($enc->decode($fn));
306 if (defined $full_pfx) {
307 $rv .= add_text_body_short($enc, $part, $part_nr,
310 $rv .= add_text_body_full($enc, $part, $part_nr);
312 $rv .= "\n" unless $rv =~ /\n\z/s;
318 sub add_filename_line {
324 $pad x= ($len/2) if ($len > 0);
325 "$pad " . ascii_html($fn) . " $pad\n";
328 my $LINK_RE = qr!\b((?:ftp|https?|nntp)://[@\w\+\&\?\.\%\;/#=-]+)!;
331 # no newlines added here since it'd break the splitting we do
333 $_[0] =~ s!$LINK_RE!<a href="$1">$1</a>!g;
336 sub add_text_body_short {
337 my ($enc, $part, $part_nr, $full_pfx) = @_;
339 my $s = ascii_html($enc->decode($part->body));
341 $s =~ s!^((?:(?:>[^\n]*)\n)+)!
343 my @lines = split(/\n/, $cur);
344 if (@lines > MAX_INLINE_QUOTED) {
345 # show a short snippet of quoted text
346 $cur = join(' ', @lines);
347 $cur =~ s/^>\s*//;
349 my @sum = split(/\s+/, $cur);
352 my $tmp = shift(@sum);
353 my $len = length($tmp) + length($cur);
354 if ($len > MAX_TRUNC_LEN) {
359 } while (@sum && length($cur) < MAX_TRUNC_LEN);
361 "> <<a\nhref=\"${full_pfx}#q${part_nr}_" . $n++ .
362 "\">$cur<\/a>>\n";
370 sub add_text_body_full {
371 my ($enc, $part, $part_nr) = @_;
373 my $s = ascii_html($enc->decode($part->body));
375 $s =~ s!^((?:(?:>[^\n]*)\n)+)!
377 my @lines = split(/\n/, $cur);
378 if (@lines > MAX_INLINE_QUOTED) {
379 "<a\nid=q${part_nr}_" . $n++ . ">$cur</a>";
387 sub headers_to_html_header {
388 my ($mime, $full_pfx, $srch) = @_;
392 foreach my $h (qw(From To Cc Subject Date)) {
393 my $v = $mime->header($h);
394 defined($v) && length($v) or next;
395 $v = PublicInbox::Hval->new_oneline($v);
398 my @from = Email::Address->parse($v->raw);
399 $title[1] = ascii_html($from[0]->name);
400 } elsif ($h eq 'Subject') {
401 $title[0] = $v->as_html;
403 my $path = $srch->subject_path($v->raw);
404 $rv .= "$h: <a\nhref=\"../s/$path.html\">";
405 $rv .= $v->as_html . "</a>\n";
409 $rv .= "$h: " . $v->as_html . "\n";
413 my $header_obj = $mime->header_obj;
414 my $mid = $header_obj->header_raw('Message-ID');
415 $mid = PublicInbox::Hval->new_msgid($mid);
416 $rv .= 'Message-ID: <' . $mid->as_html . '> ';
417 my $href = $mid->as_href;
418 $href = "../m/$href" unless $full_pfx;
419 $rv .= "(<a\nhref=\"$href.txt\">raw</a>)\n";
421 my $irt = $header_obj->header_raw('In-Reply-To');
423 my $v = PublicInbox::Hval->new_msgid($irt);
424 my $html = $v->as_html;
425 my $href = $v->as_href;
426 $rv .= "In-Reply-To: <";
427 $rv .= "<a\nhref=\"$href.html\">$html</a>>\n";
430 my $refs = $header_obj->header_raw('References');
432 $refs =~ s/\s*\Q$irt\E\s*// if (defined $irt);
433 my @refs = ($refs =~ /<([^>]+)>/g);
435 $rv .= 'References: '. linkify_refs(@refs) . "\n";
441 ("<html><head><title>". join(' - ', @title) .
442 '</title></head><body>' . PRE_WRAP . $rv);
446 my ($mime, $standalone, $full_pfx, $srch) = @_;
447 my %cc; # everyone else
448 my $to; # this is the From address
450 foreach my $h (qw(From To Cc)) {
451 my $v = $mime->header($h);
452 defined($v) && length($v) or next;
453 my @addrs = Email::Address->parse($v);
454 foreach my $recip (@addrs) {
455 my $address = $recip->address;
456 my $dst = lc($address);
457 $cc{$dst} ||= $address;
461 Email::Address->purge_cache if $standalone;
463 my $subj = $mime->header('Subject') || '';
464 $subj = "Re: $subj" unless $subj =~ /\bRe:/;
465 my $mid = $mime->header_obj->header_raw('Message-ID');
466 my $irt = uri_escape_utf8($mid);
468 $to = uri_escape_utf8($to);
469 $subj = uri_escape_utf8($subj);
471 my $cc = uri_escape_utf8(join(',', sort values %cc));
472 my $href = "mailto:$to?In-Reply-To=$irt&Cc=${cc}&Subject=$subj";
474 my $idx = $standalone ? " <a\nhref=\"../\">index</a>" : '';
476 $irt = $mime->header_obj->header_raw('In-Reply-To') || '';
477 $mid = mid_compressed(mid_clean($mid));
478 my $t_anchor = length $irt ? T_ANCHOR : '';
479 $idx = " <a\nhref=\"../t/$mid.html$t_anchor\">thread</a>$idx";
480 my $res = $srch->get_replies($mid);
481 if (my $c = $res->{count}) {
482 $c = $c == 1 ? '1 reply' : "$c replies";
484 thread_replies(\$idx, $mime, $res);
486 $idx .= "\n(no replies yet)\n";
489 $irt = PublicInbox::Hval->new_msgid($irt);
490 $irt = $irt->as_href;
491 $irt = "<a\nhref=\"$irt\">parent</a> ";
493 $irt = ' ' x length('parent ');
499 "$irt<a\nhref=\"" . ascii_html($href) . '">reply</a>' . $idx;
504 my $v = PublicInbox::Hval->new_msgid($_);
505 my $html = $v->as_html;
506 my $href = $v->as_href;
507 "<<a\nhref=\"$href.html\">$html</a>>";
513 'm' . mid_compressed(mid_clean($msgid));
517 my ($dst, $root, $node, $level) = @_;
518 my $pfx = ' ' x $level;
520 if (my $x = $node->message) {
521 my $mid = $x->header('Message-ID');
522 if ($root->[0] ne $mid) {
523 my $s = $x->header('Subject');
524 my $h = hash_subj($s);
525 if ($root->[1]->{$h}) {
528 $root->[1]->{$h} = 1;
529 $s = PublicInbox::Hval->new($s);
532 my $m = PublicInbox::Hval->new_msgid($mid);
533 my $f = PublicInbox::Hval->new($x->header('X-PI-From'));
534 my $d = PublicInbox::Hval->new($x->header('X-PI-Date'));
535 $m = $m->as_href . '.html';
537 $d = $d->as_html . ' UTC';
538 if (length($s) == 0) {
539 $$dst .= "` <a\nhref=\"$m\">$f @ $d</a>\n";
541 $$dst .= "` <a\nhref=\"$m\">$s</a>\n" .
546 simple_dump($dst, $root, $node->child, $level + 1) if $node->child;
547 simple_dump($dst, $root, $node->next, $level) if $node->next;
554 $subj =~ s/^(?:re|aw):\s*//i; # remove reply prefix (aw: German)
556 Digest::SHA::sha1($subj);
560 my ($dst, $root, $res) = @_;
561 my @msgs = map { $_->mini_mime } @{$res->{msgs}};
562 foreach (@{$res->{msgs}}) {
563 print STDERR "smsg->path: <", $_->path, ">\n";
565 require PublicInbox::Thread;
566 $root->header_set('X-PI-TS', '0');
567 my $th = PublicInbox::Thread->new($root, @msgs);
569 $th->order(*PublicInbox::Thread::sort_ts);
570 $root = [ $root->header('Message-ID'),
571 { hash_subj($root->header('Subject')) => 1 } ];
572 simple_dump($dst, $root, $_, 0) for $th->rootset;
575 sub thread_html_head {
577 my $s = PublicInbox::Hval->new_oneline($mime->header('Subject'));
579 "<html><head><title>$s</title></head><body>" . PRE_WRAP
584 my ($dst, $state, $node, $level) = @_;
585 # $state = [ $search_res, $seen, undef, 0 (msg_nr) ];
586 # $seen is overloaded with 3 types of fields:
587 # 1) "root_anchor" => anchor_for(Message-ID),
588 # 2) seen subject hashes: sha1(subject) => 1
589 # 3) anchors hashes: "#$sha1_hex" (same as $seen in index_entry)
590 if (my $mime = $node->message) {
591 if (length($$dst) == 0) {
592 $$dst .= thread_html_head($mime);
594 $$dst .= index_entry(undef, $mime, $level, $state);
596 thread_entry($dst, $state, $node->child, $level + 1) if $node->child;
597 thread_entry($dst, $state, $node->next, $level) if $node->next;