dist/ntp/ntp-html2mdoc.pl

   1 #!/usr/pkg/bin/perl
   2
   3 @longmonth = ( "January", "February", "March", "April", "May", "June", "July",
   4                "August", "September", "October", "November", "December" );
   5 # require "ctime.pl";
   6
   7 $sec = 8;                               # XXX
   8
   9 sub swallow {
  10         my ($what) = @_;
  11
  12         while (<>) {
  13                 chomp;
  14                 return if /$what/i;
  15         }
  16 }
  17
  18 sub chew {
  19         my ($match, $reset) = @_;
  20
  21         $_ = undef if $reset;
  22         while (!(/$match/i)) {
  23                 chomp($_ .= " " . <>);
  24         }
  25 }
  26
  27 sub detag {
  28         s/<[^>]+>//g;
  29 }
  30
  31 sub dehtmlchar {
  32         s/&nbsp;/ /g;
  33         s/&lt;/</g;
  34         s/&gt;/>/g;
  35 }
  36
  37 ($mday, $mon, $year) = (localtime(time))[3 .. 5];
  38 $date = sprintf "%s %02d, %d", $longmonth[$mon], $mday, $year + 1900;
  39
  40 swallow("<H3>");
  41 chomp($_ = <>);
  42 detag;
  43 ($name, $descr) = split(' ', $_, 2);
  44 $descr =~ s/^\s*-\s*//;
  45 ($NAME = $name) =~ tr/a-z/A-Z/;
  46
  47 print <<EOF;
  48 .\\"    \$NetBSD\$
  49 .\\" Converted from HTML to mandoc by ntp-html2mdoc.pl
  50 .\\"
  51 .Dd $date
  52 .Dt $NAME $sec
  53 .Os
  54 .Sh NAME
  55 .Nm $name
  56 .Nd $descr
  57 .Sh SYNOPSIS
  58 .Nm
  59 EOF
  60
  61 swallow("Synopsis");
  62 chew("</TT>", 1);
  63 detag;
  64 s/^\s*$name\s*//;
  65 @args = split;
  66 while ($_ = shift @args) {
  67         next if /\[/;
  68         s/^-//;
  69         print ".Op Fl $_";
  70         if (defined($args[0]) && $args[0] ne "]") {
  71                 print " Ar ", shift @args;
  72         }
  73         print "\n";
  74         shift @args;    # remove the "]"
  75 }
  76
  77 while (<>) {
  78         chomp;
  79         next if length == 0;
  80         if (/^<H4>/i) {
  81                 chew("</H4>");
  82                 detag;
  83                 tr/a-z/A-Z/;
  84                 s/^\s+//;
  85                 dehtmlchar;
  86                 print ".Sh $_\n";
  87                 next;
  88         }
  89         if (/<ADDRESS>/i) {
  90                 chew("</ADDRESS>");
  91                 detag;
  92                 s/^\s+//;
  93                 dehtmlchar;
  94                 print ".Sh AUTHOR\n$_\n";
  95                 next;
  96         }
  97         if (/<DL>/i) {
  98                 print ".Bl -tag -width indent\n";
  99                 next;
 100         }
 101         if (m#</DL>#i) {
 102                 print ".El\n";
 103                 next;
 104         }
 105         if (/<DT>/i) {
 106                 chew("</DT>");
 107                 detag;
 108                 s/^\s*-//;
 109                 s/ / Ar /;
 110                 dehtmlchar;
 111                 print ".It Fl $_\n";
 112                 next;
 113         }
 114         if (/<TT>-/) {
 115                 # command line option
 116                 chew("</TT>");
 117                 s#<TT>-([^<]*)#\n.Fl $1\n#ig;
 118                 s#</TT>##ig;
 119                 s#<I>([^<]+)</I>#\n.Ar $1\n#ig;
 120         }
 121         if (/<A HREF/) {
 122                 # html reference to another ntp page
 123                 chew("</A>");
 124                 chomp($_ .= " " . <>) if (/<A$/);       # another reference on the next line
 125                 s#<TT><A HREF="(.*).htm">\1</A></TT>#\n.Xr \1 $sec\n#ig;
 126                 s#<A HREF="([^"]*)">(.*?) +</A> *page#\n.%T "$2"\npage in\n.Pa /usr/share/doc/html/ntp/$1\n#ig;
 127                 s#^<BR>##g;
 128                 s#<A HREF="([^"]*)">(.*?)</A>#For\n.%T "$2"\n, refer to\n.Pa /usr/share/doc/html/ntp/$1 .\n.Pp\n#ig;
 129                 detag;
 130         }
 131         s#<TT>$name</TT>#\n.Nm\n#ig;
 132         s#<TT>([^<]*)</TT>#\n.Pa $1\n#ig;
 133         s#<TT><A HREF=[^>]*>([^<]*)</A></TT>#\n.Pa $1\n#ig;
 134         s#<PRE>#.Pp\n.nf\n#ig;
 135         s#</PRE>#\n.fi\n.Pp\n#ig;
 136         s#<P>#\n.Pp\n#ig;
 137         s#<(HR|DD)>##ig;
 138         s#</(BODY|HTML|DD)>##ig;
 139         s#<I>([^<]+)</I>#\n.Ar $1\n#ig;
 140         dehtmlchar;
 141         s/^\s+//;
 142         s/\n+\s*/\n/g;
 143         s/\n\.(Pa|%T) (.*)\n([,\.:]) /\n.$1 $2 $3\n/g;
 144         s/\s+\n/\n/g;
 145         s/\n$//;
 146         print $_, "\n" if length > 0;
 147 }