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