Documentation/txt2pre

   1 #!perl -w
   2 # n.b. this is invoked via $(PERL) in makefiles
   3 # Copyright (C) all contributors <meta@public-inbox.org>
   4 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
   5 #
   6 # Stupid script to make HTML from preformatted, utf-8 text versions,
   7 # only generating links for http(s).  Markdown does too much
   8 # and requires indentation to output preformatted text.
   9 use v5.12;
  10 use PublicInbox::Linkify;
  11 use PublicInbox::Hval qw(ascii_html);
  12 my (%xurls, %lei);
  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-reindex(1)
  46         lei-rm(1)
  47         lei-rm-watch(1)
  48         lei-security(7)
  49         lei-store-format(5)
  50         lei-tag(1)
  51         lei-up(1)
  52         public-inbox.cgi(1)
  53         public-inbox-clone(1)
  54         public-inbox-config(5)
  55         public-inbox-config(5)
  56         public-inbox-convert(1)
  57         public-inbox-daemon(8)
  58         public-inbox-edit(1)
  59         public-inbox-fetch(1)
  60         public-inbox-glossary(7)
  61         public-inbox-httpd(1)
  62         public-inbox-imapd(1)
  63         public-inbox-index(1)
  64         public-inbox-init(1)
  65         public-inbox-learn(1)
  66         public-inbox-mda(1)
  67         public-inbox-netd(1)
  68         public-inbox-nntpd(1)
  69         public-inbox-overview(7)
  70         public-inbox-pop3d(1)
  71         public-inbox-purge(1)
  72         public-inbox-v1-format(5)
  73         public-inbox-v2-format(5)
  74         public-inbox-watch(1)
  75         public-inbox-xcpdb(1)
  76 ]) {
  77         my ($n) = (/([\w\-\.]+)/);
  78         $xurls{$_} = "$n.html";
  79         $xurls{$n} = "$n.html";
  80         /\Alei-(.+?)\(1\)\z/ and $xurls{"lei $1"} = "$n.html";
  81 }
  82
  83 $xurls{'lei/store'} = 'lei-store-format.html';
  84
  85 for (qw[make(1) flock(2) setrlimit(2) vfork(2) tmpfs(5) inotify(7) unix(7)
  86                 syslog(3)]) {
  87         my ($n, $s) = (/([\w\-]+)\((\d)\)/);
  88         $xurls{$_} = "https://www.man7.org/linux/man-pages/man$s/$n.$s.html";
  89 }
  90
  91 for (qw[git(1)
  92         git-am(1)
  93         git-apply(1)
  94         git-config(1)
  95         git-credential(1)
  96         git-daemon(1)
  97         git-diff(1)
  98         git-fast-import(1)
  99         git-fetch(1)
 100         git-filter-branch(1)
 101         git-format-patch(1)
 102         git-gc(1)
 103         git-http-backend(1)
 104         git-imap-send(1)
 105         git-init(1)
 106         git-send-email(1)
 107         gitrepository-layout(5)
 108         gitglossary(7)
 109 ]) {
 110         my ($n) = (/([\w\-\.]+)/);
 111         $xurls{$_} = "https://kernel.org/pub/software/scm/git/docs/$n.html"
 112 }
 113
 114 for (qw[
 115         sd_listen_fds(3)
 116         systemd(1)
 117         systemd.unit(5)
 118         systemd.socket(5)
 119 ]) {
 120         my ($n) = (/([\w\-\.]+)/);
 121         $xurls{$_} = "https://www.freedesktop.org/software/systemd/man/$n.html";
 122 }
 123
 124 # favor upstream docs if they exist, use manpages.debian.org if they don't
 125 $xurls{'netrc(5)'} = 'https://manpages.debian.org/stable/ftp/netrc.5.en.html';
 126 $xurls{'mbsync(1)'} =
 127         'https://manpages.debian.org/stable/isync/mbsync.1.en.html';
 128 $xurls{'offlineimap(1)'} =
 129         'https://manpages.debian.org/stable/offlineimap/offlineimap.1.en.html';
 130 $xurls{'spamc(1)'} =
 131         'https://spamassassin.apache.org/full/3.4.x/doc/spamc.html';
 132 $xurls{'grok-pull'} =
 133         'https://git.kernel.org/pub/scm/utils/grokmirror/grokmirror.git' .
 134         '/tree/man/grok-pull.1.rst';
 135 $xurls{'git-filter-repo(1)'} = 'https://github.com/newren/git-filter-repo'.
 136                         '/blob/master/Documentation/git-filter-repo.txt';
 137 $xurls{'ssoma(1)'} = 'https://ssoma.public-inbox.org/ssoma.txt';
 138 $xurls{'cgitrc(5)'} = 'https://git.zx2c4.com/cgit/tree/cgitrc.5.txt';
 139 $xurls{'prove(1)'} = 'https://perldoc.perl.org/prove.html';
 140 $xurls{'mbox(5)'} = 'https://manpages.debian.org/stable/mutt/mbox.5.en.html';
 141 $xurls{'mmdf(5)'} = 'https://manpages.debian.org/stable/mutt/mmdf.5.en.html';
 142 $xurls{'mutt(1)'} = 'https://manpages.debian.org/stable/mutt/mutt.1.en.html';
 143 $xurls{'torsocks(1)'} =
 144         'https://manpages.debian.org/stable/torsocks/torsocks.1.en.html';
 145 $xurls{'curl(1)'} = 'https://manpages.debian.org/stable/curl/curl.1.en.html';
 146 $xurls{'copydatabase(1)'} =
 147  'https://manpages.debian.org/stable/xapian-tools/copydatabase.1.en.html';
 148 $xurls{'xapian-compact(1)'} =
 149  'https://manpages.debian.org/stable/xapian-tools/xapian-compact.1.en.html';
 150 $xurls{'gzip(1)'} = 'https://manpages.debian.org/stable/gzip/gzip.1.en.html';
 151 $xurls{'chmod(1)'} =
 152         'https://manpages.debian.org/stable/coreutils/chmod.1.en.html';
 153 $xurls{'kqueue(2)'} =
 154         'https://www.freebsd.org/cgi/man.cgi?query=kqueue&sektion=2';
 155 $xurls{'notmuch(1)'} = 'https://notmuchmail.org/manpages/notmuch-1/';
 156 $xurls{'mairix(1)'} =
 157         'https://manpages.debian.org/stable/mairix/mairix.1.en.html';
 158
 159 my $str = do { local $/; <STDIN> };
 160 my ($title) = ($str =~ /\A([^\n]+)/);
 161 if ($str =~ /^NAME\n\s+([^\n]+)/sm) {
 162         # don't link to ourselves
 163         $title = $1;
 164         if ($title =~ /([\w\.\-]+)/) {
 165                 delete $xurls{$1};
 166         }
 167         if ($title =~ /\blei-([\w\-]+)\b/) {
 168                 delete $xurls{"lei $1"};
 169         }
 170 }
 171 $title = ascii_html($title);
 172 my $l = PublicInbox::Linkify->new;
 173 $str = $l->linkify_1($str);
 174 $str = ascii_html($str);
 175
 176 # longest matches, first
 177 my @keys = sort { length($b) <=> length($a) } keys %xurls;
 178 my $xkeys = join('|', map { quotemeta } @keys);
 179 $str =~ s,(?<![>\w_])($xkeys)(?!(?:[\w<\-]|\.html)),
 180         qq(<a\nhref=").$xurls{$1}.qq(">$1).($2//'').'</a>',sge;
 181
 182 $str = $l->linkify_2($str);
 183
 184 print '<html><head>',
 185   qq(<meta\nhttp-equiv="Content-Type"\ncontent="text/html; charset=utf-8"\n/>),
 186   "<title>$title</title>",
 187   "</head><body><pre>",  $str , '</pre></body></html>';
 188 STDOUT->flush;