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-add-watch(1)
  16         lei-blob(1)
  17         lei-config(1)
  18         lei-convert(1)
  19         lei-daemon(8)
  20         lei-daemon-kill(1)
  21         lei-daemon-pid(1)
  22         lei-edit-search(1)
  23         lei-export-kw(1)
  24         lei-forget-external(1)
  25         lei-forget-mail-sync(1)
  26         lei-forget-search(1)
  27         lei-import(1)
  28         lei-index(1)
  29         lei-init(1)
  30         lei-inspect(1)
  31         lei-lcat(1)
  32         lei-ls-external(1)
  33         lei-ls-label(1)
  34         lei-ls-mail-source(1)
  35         lei-ls-mail-sync(1)
  36         lei-ls-search(1)
  37         lei-ls-watch(1)
  38         lei-mail-diff(1)
  39         lei-mail-sync-overview(7)
  40         lei-overview(7)
  41         lei-p2q(1)
  42         lei-q(1)
  43         lei-rediff(1)
  44         lei-refresh-mail-sync(1)
  45         lei-rm(1)
  46         lei-rm-watch(1)
  47         lei-security(7)
  48         lei-store-format(5)
  49         lei-tag(1)
  50         lei-up(1)
  51         public-inbox.cgi(1)
  52         public-inbox-clone(1)
  53         public-inbox-config(5)
  54         public-inbox-config(5)
  55         public-inbox-convert(1)
  56         public-inbox-daemon(8)
  57         public-inbox-edit(1)
  58         public-inbox-fetch(1)
  59         public-inbox-glossary(7)
  60         public-inbox-httpd(1)
  61         public-inbox-imapd(1)
  62         public-inbox-index(1)
  63         public-inbox-init(1)
  64         public-inbox-learn(1)
  65         public-inbox-mda(1)
  66         public-inbox-nntpd(1)
  67         public-inbox-overview(7)
  68         public-inbox-purge(1)
  69         public-inbox-v1-format(5)
  70         public-inbox-v2-format(5)
  71         public-inbox-watch(1)
  72         public-inbox-xcpdb(1)
  73 ]) {
  74         my ($n) = (/([\w\-\.]+)/);
  75         $xurls{$_} = "$n.html";
  76         $xurls{$n} = "$n.html";
  77 }
  78
  79 for (qw[make(1) flock(2) setrlimit(2) vfork(2) tmpfs(5)]) {
  80         my ($n, $s) = (/([\w\-]+)\((\d)\)/);
  81         $xurls{$_} = "http://www.man7.org/linux/man-pages/man$s/$n.$s.html";
  82 }
  83
  84 for (qw[git(1)
  85         git-am(1)
  86         git-apply(1)
  87         git-config(1)
  88         git-credential(1)
  89         git-daemon(1)
  90         git-diff(1)
  91         git-fast-import(1)
  92         git-fetch(1)
  93         git-filter-branch(1)
  94         git-format-patch(1)
  95         git-gc(1)
  96         git-http-backend(1)
  97         git-imap-send(1)
  98         git-init(1)
  99         git-send-email(1)
 100         gitrepository-layout(5)
 101         gitglossary(7)
 102 ]) {
 103         my ($n) = (/([\w\-\.]+)/);
 104         $xurls{$_} = "https://kernel.org/pub/software/scm/git/docs/$n.html"
 105 }
 106
 107 for (qw[
 108         sd_listen_fds(3)
 109         systemd(1)
 110         systemd.unit(5)
 111         systemd.socket(5)
 112 ]) {
 113         my ($n) = (/([\w\-\.]+)/);
 114         $xurls{$_} = "https://www.freedesktop.org/software/systemd/man/$n.html";
 115 }
 116
 117 # favor upstream docs if they exist, use manpages.debian.org if they don't
 118 $xurls{'netrc(5)'} = 'https://manpages.debian.org/stable/ftp/netrc.5.en.html';
 119 $xurls{'mbsync(1)'} =
 120         'https://manpages.debian.org/stable/isync/mbsync.1.en.html';
 121 $xurls{'offlineimap(1)'} =
 122         'https://manpages.debian.org/stable/offlineimap/offlineimap.1.en.html';
 123 $xurls{'spamc(1)'} =
 124         'https://spamassassin.apache.org/full/3.4.x/doc/spamc.html';
 125 $xurls{'grok-pull'} =
 126         'https://git.kernel.org/pub/scm/utils/grokmirror/grokmirror.git' .
 127         '/tree/man/grok-pull.1.rst';
 128 $xurls{'git-filter-repo(1)'} = 'https://github.com/newren/git-filter-repo'.
 129                         '/blob/master/Documentation/git-filter-repo.txt';
 130 $xurls{'ssoma(1)'} = 'https://ssoma.public-inbox.org/ssoma.txt';
 131 $xurls{'cgitrc(5)'} = 'https://git.zx2c4.com/cgit/tree/cgitrc.5.txt';
 132 $xurls{'prove(1)'} = 'https://perldoc.perl.org/prove.html';
 133 $xurls{'mbox(5)'} = 'https://manpages.debian.org/stable/mutt/mbox.5.en.html';
 134 $xurls{'mmdf(5)'} = 'https://manpages.debian.org/stable/mutt/mmdf.5.en.html';
 135 $xurls{'torsocks(1)'} =
 136         'https://manpages.debian.org/stable/torsocks/torsocks.1.en.html';
 137 $xurls{'curl(1)'} = 'https://manpages.debian.org/stable/curl/curl.1.en.html';
 138 $xurls{'copydatabase(1)'} =
 139  'https://manpages.debian.org/stable/xapian-tools/copydatabase.1.en.html';
 140 $xurls{'xapian-compact(1)'} =
 141  'https://manpages.debian.org/stable/xapian-tools/xapian-compact.1.en.html';
 142
 143 my $str = do { local $/; <STDIN> };
 144 my ($title) = ($str =~ /\A([^\n]+)/);
 145 if ($str =~ /^NAME\n\s+([^\n]+)/sm) {
 146         # don't link to ourselves
 147         $title = $1;
 148         if ($title =~ /([\w\.\-]+)/) {
 149                 delete $xurls{$1};
 150         }
 151 }
 152 $title = ascii_html($title);
 153 my $l = PublicInbox::Linkify->new;
 154 $str = $l->linkify_1($str);
 155 $str = ascii_html($str);
 156
 157 # longest matches, first
 158 my @keys = sort { length($b) <=> length($a) } keys %xurls;
 159 my $xkeys = join('|', map { quotemeta } @keys);
 160 $str =~ s,(?<![>\w_])($xkeys)(?!(?:[\w<\-]|\.html)),
 161         qq(<a\nhref=").$xurls{$1}.qq(">$1).($2//'').'</a>',sge;
 162
 163 $str = $l->linkify_2($str);
 164
 165 print '<html><head>',
 166   qq(<meta\nhttp-equiv="Content-Type"\ncontent="text/html; charset=utf-8"\n/>),
 167   "<title>$title</title>",
 168   "</head><body><pre>",  $str , '</pre></body></html>';
 169 STDOUT->flush;
 170
 171 # keep mtime on website consistent so clients can cache
 172 if (-f STDIN && -f STDOUT) {
 173         my @st = stat(STDIN);
 174         utime($st[8], $st[9], \*STDOUT);
 175 }