#! /usr/bin/perl

# $Cambridge: exim/doc/doc-docbook/Pre-xml,v 1.1 2005/06/16 10:32:31 ph10 Exp $

# Script to pre-process XML input before processing it for various purposes.
# Options specify which transformations are to be done. Monospaced literal
# layout blocks are never touched.

# Changes:

# -abstract: Remove the <abstract> element

# -ascii:    Replace &8230;   (sic, no x) with ...
#            Replace &#x2019; by '
#            Replace &#x201C; by "
#            Replace &#x201D; by "
#            Replace &#x2013; by -
#            Replace &#x2020; by *
#            Replace &#x2021; by **
#            Replace &#x00a0; by a space
#            Replace &#169;   by (c)
#            Put quotes round <literal> text
#            Put quotes round <quote> text

# -bookinfo: Remove the <bookinfo> element from the file

# -fi:       Replace "fi" by &#xFB01; except when it is in an XML element, or
#            inside a <literal>.

# -noindex   Remove the XML to generate a Concept and an Options index.
# -oneindex  Ditto, but add XML to generate a single index.



# The function that processes non-literal monospaced text

sub process()
{
my($s) = $_[0];

$s =~ s/fi(?![^<>]*>)/&#xFB01;/g if $ligatures;

if ($ascii)
  {
  $s =~ s/&#8230;/.../g;
  $s =~ s/&#x2019;/'/g;
  $s =~ s/&#x201C;/"/g;
  $s =~ s/&#x201D;/"/g;
  $s =~ s/&#x2013;/-/g;
  $s =~ s/&#x2020;/*/g;
  $s =~ s/&#x2021;/**/g;
  $s =~ s/&#x00a0;/ /g;
  $s =~ s/&#x00a9;/(c)/g;
  $s =~ s/<quote>/"/g;
  $s =~ s/<\/quote>/"/g;
  }

$s;
}


# The main program

$abstract  = 0;
$ascii     = 0;
$bookinfo  = 0;
$inliteral = 0;
$ligatures = 0;
$madeindex = 0;
$noindex   = 0;
$oneindex  = 0;

foreach $arg (@ARGV)
  {
  if    ($arg eq "-fi")       { $ligatures = 1; }
  elsif ($arg eq "-abstract") { $abstract = 1; }
  elsif ($arg eq "-ascii")    { $ascii = 1; }
  elsif ($arg eq "-bookinfo") { $bookinfo = 1; }
  elsif ($arg eq "-noindex")  { $noindex = 1; }
  elsif ($arg eq "-oneindex") { $oneindex = 1; }
  else  { die "** Pre-xml: Unknown option \"$arg\"\n"; }
  }

while (<STDIN>)
  {
  # Remove <abstract> if required

  next if ($abstract && /^\s*<abstract>/);

  # Remove <bookinfo> if required

  if ($bookinfo && /^<bookinfo/)
    {
    while (<STDIN>) { last if /^<\/bookinfo/; }
    next;
    }

  # Copy monospaced literallayout blocks

  if (/^<literallayout class="monospaced">/)
    {
    print;
    while (<STDIN>)
      {
      print;
      last if /^<\/literallayout>/;
      }
    next;
    }

  # Adjust index-generation code if required

  if (($noindex || $oneindex) && /^<index[\s>]/)
    {
    while (<STDIN>)
      {
      last if /^<\/index>/;
      }

    if ($oneindex && !$madeindex)
      {
      $madeindex = 1;
      print "<index><title>Index</title></index>\n";
      }

    next;
    }

  # A line that is not in a monospaced literal block; keep track of which
  # parts are in <literal> and which not. The latter get processed by the
  # function above.

  for (;;)
    {
    if ($inliteral)
      {
      if (/^(.*?)<\/literal>(.*)$/)
        {
        print $1;
        print "\"" if $ascii;
        print "</literal>";
        $inliteral = 0;
        $_ = "$2\n";
        }
      else
        {
        print;
        last;
        }
      }

    # Not in literal state

    else
      {
      if (/^(.*?)<literal>(.*)$/)
        {
        print &process($1);
        print "<literal>";
        print "\"" if $ascii;
        $inliteral = 1;
        $_ = "$2\n";
        }
      else
        {
        print &process($_);
        last;
        }
      }
    }    # Loop for different parts of one line
  }      # Loop for multiple lines

# End