Documentation/txt2pre

   1 #!/usr/bin/env perl
   2 # Copyright (C) 2014-2021 all contributors <meta@public-inbox.org>
   3 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
   4 #
   5 # Stupid script to make HTML from preformatted, utf-8 text versions,
   6 # only generating links for http(s).  Markdown does too much
   7 # and requires indentation to output preformatted text.
   8 use strict;
   9 use warnings;
  10 use PublicInbox::Linkify;
  11 use PublicInbox::Hval qw(ascii_html);
  12 my %xurls;
  13 for (qw[lei(1)
  14         lei-add-external(1)
  15         lei-blob(1)
  16         lei-config(1)
  17         lei-daemon-kill(1)
  18         lei-daemon-pid(1)
  19         lei-forget-external(1)
  20         lei-import(1)
  21         lei-init(1)
  22         lei-ls-external(1)
  23         lei-ls-label(1)
  24         lei-tag(1)
  25         lei-overview(7)
  26         lei-p2q(1)
  27         lei-q(1)
  28         public-inbox.cgi(1)
  29         public-inbox-compact(1)
  30         public-inbox-config(5)
  31         public-inbox-convert(1)
  32         public-inbox-daemon(8)
  33         public-inbox-edit(1)
  34         public-inbox-glossary(7)
  35         public-inbox-httpd(1)
  36         public-inbox-imapd(1)
  37         public-inbox-index(1)
  38         public-inbox-init(1)
  39         public-inbox-learn(1)
  40         public-inbox-mda(1)
  41         public-inbox-nntpd(1)
  42         public-inbox-overview(7)
  43         public-inbox-purge(1)
  44         public-inbox-v1-format(5)
  45         public-inbox-v2-format(5)
  46         public-inbox-watch(1)
  47         public-inbox-xcpdb(1)
  48 ]) {
  49         my ($n) = (/([\w\-\.]+)/);
  50         $xurls{$_} = "$n.html";
  51         $xurls{$n} = "$n.html";
  52 }
  53
  54 for (qw[copydatabase(1) xapian-compact(1)]) {
  55         my ($n) = (/([\w\-\.]+)/);
  56         $xurls{$_} = ".$n.1.html"
  57 }
  58
  59 for (qw[make(1) flock(2) setrlimit(2) vfork(2) tmpfs(5)]) {
  60         my ($n, $s) = (/([\w\-]+)\((\d)\)/);
  61         $xurls{$_} = "http://www.man7.org/linux/man-pages/man$s/$n.$s.html";
  62 }
  63
  64 for (qw[git(1)
  65         git-am(1)
  66         git-apply(1)
  67         git-config(1)
  68         git-credential(1)
  69         git-daemon(1)
  70         git-fast-import(1)
  71         git-fetch(1)
  72         git-filter-branch(1)
  73         git-format-patch(1)
  74         git-gc(1)
  75         git-http-backend(1)
  76         git-imap-send(1)
  77         git-init(1)
  78         git-send-email(1)
  79         gitrepository-layout(5)
  80 ]) {
  81         my ($n) = (/([\w\-\.]+)/);
  82         $xurls{$_} = "https://kernel.org/pub/software/scm/git/docs/$n.html"
  83 }
  84
  85 for (qw[
  86         sd_listen_fds(3)
  87         systemd(1)
  88         systemd.unit(5)
  89         systemd.socket(5)
  90 ]) {
  91         my ($n) = (/([\w\-\.]+)/);
  92         $xurls{$_} = "https://www.freedesktop.org/software/systemd/man/$n.html";
  93 }
  94
  95 $xurls{'netrc(5)'} = 'https://manpages.debian.org/stable/ftp/netrc.5.en.html';
  96 $xurls{'mbsync(1)'} =
  97         'https://manpages.debian.org/stable/isync/mbsync.1.en.html';
  98 $xurls{'offlineimap(1)'} =
  99         'https://manpages.debian.org/stable/offlineimap/offlineimap.1.en.html';
 100 $xurls{'spamc(1)'} =
 101         'https://spamassassin.apache.org/full/3.4.x/doc/spamc.html';
 102 $xurls{'grok-pull'} =
 103         'https://git.kernel.org/pub/scm/utils/grokmirror/grokmirror.git' .
 104         '/tree/man/grok-pull.1.rst';
 105 $xurls{'git-filter-repo(1)'} = 'https://github.com/newren/git-filter-repo'.
 106                         '/blob/master/Documentation/git-filter-repo.txt';
 107 $xurls{'ssoma(1)'} = 'https://ssoma.public-inbox.org/ssoma.txt';
 108 $xurls{'cgitrc(5)'} = 'https://git.zx2c4.com/cgit/tree/cgitrc.5.txt';
 109 $xurls{'prove(1)'} = 'https://perldoc.perl.org/prove.html';
 110 $xurls{'mbox(5)'} = 'https://manpages.debian.org/stable/mutt/mbox.5.en.html';
 111
 112 my $str = do { local $/; <STDIN> };
 113 my ($title) = ($str =~ /\A([^\n]+)/);
 114 if ($str =~ /^NAME\n\s+([^\n]+)/sm) {
 115         # don't link to ourselves
 116         $title = $1;
 117         if ($title =~ /([\w\.\-]+)/) {
 118                 delete $xurls{$1};
 119         }
 120 }
 121 $title = ascii_html($title);
 122 my $l = PublicInbox::Linkify->new;
 123 $str = $l->linkify_1($str);
 124 $str = ascii_html($str);
 125
 126 # longest matches, first
 127 my @keys = sort { length($b) <=> length($a) } keys %xurls;
 128 my $xkeys = join('|', map { quotemeta } @keys);
 129 $str =~ s,(?<![>\w_])($xkeys)(?!(?:[\w<\-]|\.html)),
 130         qq(<a\nhref=").$xurls{$1}.qq(">$1).($2//'').'</a>',sge;
 131
 132 $str = $l->linkify_2($str);
 133
 134 print '<html><head>',
 135   qq(<meta\nhttp-equiv="Content-Type"\ncontent="text/html; charset=utf-8"\n/>),
 136   "<title>$title</title>",
 137   "</head><body><pre>",  $str , '</pre></body></html>';
 138 STDOUT->flush;
 139
 140 # keep mtime on website consistent so clients can cache
 141 if (-f STDIN && -f STDOUT) {
 142         my @st = stat(STDIN);
 143         utime($st[8], $st[9], \*STDOUT);
 144 }