1 7d6d378a 2022-08-24 op #!/usr/bin/env perl
3 1bcb9899 2022-08-29 op # mkindex was written by Omar Polo <op@openbsd.org> and is placed in
4 1bcb9899 2022-08-29 op # the public domain. The author hereby disclaims copyright to this
7 7d6d378a 2022-08-24 op use open ":std", ":encoding(UTF-8)";
12 270695a3 2022-08-30 op use File::Temp qw(tempfile);
14 054f3fd4 2022-08-25 op use OpenBSD::Pledge;
15 054f3fd4 2022-08-25 op use OpenBSD::Unveil;
17 bd3babdf 2023-04-02 op use GotMArc qw(parse san urlencode initpage endpage index_header
18 789a106a 2023-05-05 op search thread_header threntry);
20 7d6d378a 2022-08-24 op my $outdir = $ENV{'OUTDIR'};
21 7d6d378a 2022-08-24 op die 'Set $OUTDIR' unless defined $outdir;
23 28ebc168 2022-08-24 op my $tfh; # thread file handle
24 28ebc168 2022-08-24 op my $pfh; # page file handle
30 584d65d8 2022-08-24 op my $threads_seen = 0;
32 584d65d8 2022-08-24 op my $last_level = 0;
34 debcbab2 2022-09-13 op my $last_date;
35 debcbab2 2022-09-13 op my $last_from;
36 debcbab2 2022-09-13 op my $last_subj;
38 c5dff871 2022-08-29 op my $threads = 0;
39 c5dff871 2022-08-29 op my $threads_per_page = 100;
42 d40f4fe2 2022-08-26 op my ($a, $b) = @_;
43 d40f4fe2 2022-08-26 op return $a unless defined $b;
44 d40f4fe2 2022-08-26 op return $a gt $b ? $a : $b;
48 d40f4fe2 2022-08-26 op my ($a, $b) = @_;
49 d40f4fe2 2022-08-26 op return $a unless defined $b;
50 d40f4fe2 2022-08-26 op return $a lt $b ? $a : $b;
53 28ebc168 2022-08-24 op sub pagename {
54 28ebc168 2022-08-24 op my $i = shift;
55 1913b2fb 2022-08-26 op return $i == 1 && "index.html" || "$i.html";
59 ebad21b1 2022-08-26 op say $pfh '</ul></div>';
61 d40f4fe2 2022-08-26 op push @pages, "$from_day - $to_day";
64 28ebc168 2022-08-24 op sub nextfile {
65 ebad21b1 2022-08-26 op endfile if defined $pfh;
69 270695a3 2022-08-30 op ($pfh, $path) = tempfile "/tmp/gotmarc.index.XXXXXXXXXX";
70 b031debf 2023-03-29 op binmode($pfh, ':utf8');
71 270695a3 2022-08-30 op push @files, $path;
72 ebad21b1 2022-08-26 op say $pfh "<div class='thread'><ul>";
76 43c49583 2022-08-27 op my ($pfh, $n) = @_;
77 43c49583 2022-08-27 op my ($first, $last) = (pagename(1), pagename($page));
78 6fde8229 2022-08-26 op my ($next, $prev) = (pagename($n+1), pagename($n-1));
80 2eb655f1 2022-08-27 op say $pfh "<nav>";
81 1913b2fb 2022-08-26 op say $pfh "<a href='$first'>First</a>" if $n > 2;
82 1913b2fb 2022-08-26 op say $pfh "<a href='$prev'>Prev</a>" if $n > 1;
83 6fde8229 2022-08-26 op say $pfh "<a href='$next'>Next</a>" if $n < $page;
84 6fde8229 2022-08-26 op say $pfh "<a href='$last'>Last</a>" if $n < $page - 1;
85 6fde8229 2022-08-26 op say $pfh "</nav>";
88 6fde8229 2022-08-26 op sub copyfrom {
89 6fde8229 2022-08-26 op my ($path, $fh) = @_;
91 6fde8229 2022-08-26 op # there are probably faster ways to do this like File::Copy,
92 6fde8229 2022-08-26 op # but it bypasses the bufio cache...
93 6fde8229 2022-08-26 op open(my $pfh, '<', $path) or die "can't open $path: $!";
94 890362cd 2022-08-30 op print $fh $_ while <$pfh>;
97 add05cb0 2022-08-27 op sub renderpages {
100 1913b2fb 2022-08-26 op for (my $i = 1; $i <= $page; $i++) {
101 66e1cf97 2022-08-27 op my $name = pagename($i);
102 270695a3 2022-08-30 op my $path = shift @files;
103 66e1cf97 2022-08-27 op my $dest = "$outdir/$name";
105 6fde8229 2022-08-26 op open(my $pfh, '>', $dest)
106 6fde8229 2022-08-26 op or die "can't open $dest for writing: $!";
108 39c49919 2023-05-05 op my $title = "page $i";
109 d40f4fe2 2022-08-26 op my $subtitle = $pages[$i-1];
111 9ec6c848 2022-08-27 op initpage($pfh, $title);
112 9ec6c848 2022-08-27 op index_header $pfh, $i, $subtitle;
113 9ec6c848 2022-08-27 op say $pfh "<main>";
115 43c49583 2022-08-27 op nav $pfh, $i if $page > 1;
117 6fde8229 2022-08-26 op copyfrom($path, $pfh);
118 43c49583 2022-08-27 op nav $pfh, $i if $page > 1;
120 e773a9f0 2022-08-24 op say $pfh "</main>";
121 e773a9f0 2022-08-24 op endpage($pfh);
124 8d17053d 2022-08-26 op unlink $path;
128 eefeacb7 2022-08-27 op sub endthread {
129 eefeacb7 2022-08-27 op say $tfh "</ul></li>" x $last_level;
130 c13d576c 2022-08-27 op say $tfh "</ul>\n</div>\n";
131 eefeacb7 2022-08-27 op endpage($tfh);
134 eefeacb7 2022-08-27 op $last_level = 0;
137 7d6d378a 2022-08-24 op sub nextthread {
138 eefeacb7 2022-08-27 op endthread if defined $tfh;
139 7d6d378a 2022-08-24 op my ($mid, $subj) = @_;
140 7d6d378a 2022-08-24 op my $dest = "$outdir/thread/$mid.html";
141 7d6d378a 2022-08-24 op open($tfh, '>', $dest) or die "can't open $dest: $!";
142 7d6d378a 2022-08-24 op initpage($tfh, $subj);
143 8f7a9a46 2023-04-04 op thread_header $tfh, ["Thread: $subj"];
144 c13d576c 2022-08-27 op print $tfh "<div class='thread'><ul class='mails'>\n";
147 debcbab2 2022-09-13 op sub index_entry {
148 debcbab2 2022-09-13 op my ($fh, $mid, $date, $from, $subj) = @_;
150 bd3babdf 2023-04-02 op # synthetic mail hash
154 bd3babdf 2023-04-02 op date => $date,
155 bd3babdf 2023-04-02 op from => $from,
156 bd3babdf 2023-04-02 op subj => $subj,
159 bd3babdf 2023-04-02 op threntry $fh, "thread", 0, 0, $mail;
162 054f3fd4 2022-08-25 op unveil($outdir, "rwc") or die "unveil $outdir: $!";
164 270695a3 2022-08-30 op # can't use tmppath because File::Temp checks wether /tmp exists.
165 66e1cf97 2022-08-27 op unveil("/tmp", "rwc") or die "unveil /tmp: $!";
167 270695a3 2022-08-30 op # fattr for File::Temp
168 270695a3 2022-08-30 op pledge("stdio rpath wpath cpath fattr") or die "pledge: $!";
173 6b36ff28 2023-04-01 op my $mail = parse $_;
175 bbdbef1a 2023-04-01 op if ($mail->{level} == 0) {
176 bbdbef1a 2023-04-01 op nextthread $mail->{mid}, $mail->{subj};
179 c5dff871 2022-08-29 op if ($threads > $threads_per_page) {
181 c5dff871 2022-08-29 op $threads = 0;
182 d40f4fe2 2022-08-26 op $to_day = undef;
183 d40f4fe2 2022-08-26 op $from_day = undef;
186 bbdbef1a 2023-04-01 op my $day = $mail->{date} =~ s/ .*//r;
187 a0b4d4b3 2022-08-27 op $to_day = mins $day, $to_day;
188 a0b4d4b3 2022-08-27 op $from_day = maxs $day, $from_day;
191 bd3babdf 2023-04-02 op $last_level = threntry $tfh, "mail", 0, $last_level, $mail;
192 584d65d8 2022-08-24 op $threads_seen = 1;
194 debcbab2 2022-09-13 op index_entry $pfh, $last_tid, $last_date, $last_from, $last_subj
195 bbdbef1a 2023-04-01 op if defined $last_tid && $mail->{level} == 0;
197 257d7c99 2022-10-04 op # `gt' on dates works because the format used allow for
198 257d7c99 2022-10-04 op # lexicographic comparisons.
199 bbdbef1a 2023-04-01 op if ($mail->{level} == 0 || $mail->{date} gt $last_date) {
200 bbdbef1a 2023-04-01 op $last_date = $mail->{date};
201 bbdbef1a 2023-04-01 op $last_from = $mail->{from};
204 bbdbef1a 2023-04-01 op if ($mail->{level} == 0) {
205 bbdbef1a 2023-04-01 op $last_tid = $mail->{mid};
206 bbdbef1a 2023-04-01 op $last_subj = $mail->{subj};
210 debcbab2 2022-09-13 op index_entry $pfh, $last_tid, $last_date, $last_from, $last_subj
211 debcbab2 2022-09-13 op if defined $last_tid;
214 ebad21b1 2022-08-26 op endthread if $threads_seen;