source/libs/zziplib/zziplib-src/docs/make-dbk.pl

   1 #! /usr/local/bin/perl
   2 # this file converts simple html text into a docbook xml variant.
   3 # The mapping of markups and links is far from perfect. But all we
   4 # want is the docbook-to-pdf converter and similar technology being
   5 # present in the world of docbook-to-anything converters.
   6
   7 use strict;
   8
   9 my %o;
  10
  11 my %file;
  12 my $F;
  13 my @order;
  14
  15 for $F (@ARGV)
  16 {
  17     if ($F =~ /^(\w+)=(.*)/)
  18     {
  19         $o{$1} = $2;
  20     }else{
  21         open F, "<$F" or next;
  22         my $T = join ("",<F>); close F;
  23         $file{$F}{text} = $T;
  24         $file{$F}{orig} = $F;
  25         push @order, $F;
  26     }
  27 }
  28
  29 $o{version} = `date` if not length $o{version};
  30
  31 for $F (keys %file)
  32 {
  33     $_ = $file{$F}{text};
  34     s{<!--VERSION-->}{ $o{version} }gse;
  35     s{</h2>(.*)}{</title>\n<subtitle>$1</subtitle>}mg;
  36     s{<h2>}{<sect1 id=\"$F\"><title>}mg;
  37     s{<[Pp]([> ])}{<para$1}mg; s{</[Pp]>}{</para>}mg;
  38     s{<pre>}{<screen>}mg; s{</pre>}{</screen>}mg;
  39     s{<h3>}{<sect2><title>}mg;
  40     s{</h3>((?:.(?!<sect2>))*.?)}{</title>$1</sect2>}sg;
  41     s{<!doctype [^<>]*>}{}sg;
  42     s{<!DOCTYPE [^<>]*>}{}sg;
  43     s{(<\w+\b[^<>]*\swidth=)(\d+\%)}{$1\"$2\"}sg;
  44     s{(<\w+\b[^<>]*\s\w+=)(\d+)}{$1\"$2\"}sg;
  45     s{&&}{\&amp\;\&amp\;}sg;
  46     s{\$\<}{\$\&lt\;}sg;
  47     s{&(\w+[\),])}{\&amp\;$1}sg;
  48     s{(</?)span(\s[^<>]*)?>}{$1."phrase$2>"}sge;
  49     s{(</?)small(\s[^<>]*)?>}{$1."note$2>"}sge;
  50     s{(</?)(b|em|i)>}{$1."emphasis>"}sge;
  51     s{(</?)(li)>}{$1."listitem>"}sge;
  52     s{(</?)(ul)>}{$1."itemizedlist>"}sge;
  53     s{(</?)(ol)>}{$1."orderedlist>"}sge;
  54     s{(</?)(dl)>}{$1."variablelist>"}sge;
  55     s{<dt\b([^<>]*)>}{"<varlistentry$1><term>"}sge;
  56     s{</dt\b([^<>]*)>}{"</term>"}sge;
  57     s{<dd\b([^<>]*)>}{"<listitem$1>"}sge;
  58     s{</dd\b([^<>]*)>}{"</listitem></varlistentry>"}sge;
  59     s{<table\b([^<>]*)>}{"<informaltable$1><tgroup cols=\"2\"><tbody>"}sge;
  60     s{</table\b([^<>]*)>}{"</tbody></tgroup></informaltable>"}sge;
  61     s{(</?)tr(\s[^<>]*)?>}{$1."row$2>"}sge;
  62     s{(</?)td(\s[^<>]*)?>}{$1."entry$2>"}sge;
  63     s{<informaltable\b[^<>]*>\s*<tgroup\b[^<>]*>\s*<tbody>
  64           \s*<row\b[^<>]*>\s*<entry\b[^<>]*>\s*<informaltable\b}
  65     {<informaltable}gsx;
  66     s{</informaltable>\s*</entry>\s*</row>
  67           \s*</tbody>\s*</tgroup>\s*</informaltable>}
  68     {</informaltable>}gsx;
  69     s{(<informaltable[^<>]*\swidth=\"100\%\")}{$1 pgwide=\"1\"}gs;
  70     s{(<tbody>\s*<row[^<>]*>\s*<entry[^<>]*\s)(width=\"50\%\")}
  71     {<colspec colwidth=\"1*\" /><colspec colwidth=\"1*\" />\n$1$2}gs;
  72
  73     s{<nobr>([\'\`]*)<tt>}{<cmdsynopsis>$1}sg;
  74     s{</tt>([\'\`]*)</nobr>}{$2</cmdsynopsis>}sg;
  75     s{<nobr><(?:tt|code)>([\`\"\'])}{<cmdsynopsis>$1}sg;
  76     s{<(?:tt|code)><nobr>([\`\"\'])}{<cmdsynopsis>$1}sg;
  77     s{([\`\"\'])</(?:tt|code)></nobr>}{$1</cmdsynopsis>}sg;
  78     s{([\`\"\'])</nobr></(?:tt|code)>}{$1</cmdsynopsis>}sg;
  79     s{(</?)tt>}{$1."constant>"}sge;
  80     s{(</?)code>}{$1."literal>"}sge;
  81     s{>([^<>]+)<br>}{><highlights>$1</highlights>}sg;
  82     s{<br>}{<br />}sg;
  83
  84     s{(</?)date>}{$1."sect1info>"}sge;
  85     s{<reference>}{<reference id=\"reference\">}s;
  86
  87     s{<a\s+href=\"((?:http|ftp|mailto):[^<>]+)\"\s*>((?:.(?!</a>))*.)</a>}
  88       { "<ulink url=\"$1\">$2</ulink>" }sge;
  89     s{<a\s+href=\"zziplib.html\#([\w_]+)\"\s*>((?:.(?!</a>))*.)</a>}
  90     { "<link linkend=\"$1\">$2</link>" }sge;
  91     s{<a\s+href=\"(zziplib.html)\"\s*>((?:.(?!</a>))*.)</a>}
  92     { "<link linkend=\"reference\">$2</link>" }sge;
  93     s{<a\s+href=\"([\w-]+[.]html)\"\s*>((?:.(?!</a>))*.)</a>}
  94     { my $K = $1; chop $K;
  95       if (not exists $file{$K}) { print STDERR "bad link $1\n"; }
  96       "<link linkend=\"$K\">$2</link>" }sge;
  97     s{<a\s+href=\"([\w-]+[.](?:h|c|am|txt))\"\s*>((?:.(?!</a>))*.)</a>}
  98     { "<ulink url=\"file:$1\">$2</ulink>" }sge;
  99     s{<a\s+href=\"([A-Z0-9]+[.][A-Z0-9]+)\"\s*>((?:.(?!</a>))*.)</a>}
 100     { "<ulink url=\"file:$1\">$2</ulink>" }sge;
 101
 102 #   s{(</?)subtitle>}{$1."para>"}ge;
 103
 104     $_ .= "</sect1>" if /<sect1[> ]/;
 105     $file{$F}{text} = $_;
 106 }
 107
 108 my $n = "\n";
 109
 110 print '<!DOCTYPE reference PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"',$n;
 111 print '       "http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd">',$n;
 112 print '<book><chapter><title>Documentation</title>',$n;
 113 for $F (@order)
 114 {
 115     print "</chapter>" if $file{$F}{text} =~ /<reference /;
 116     print $file{$F}{text},$n,$n;
 117 }
 118 print '</book>',$n;