#!/usr/bin/perl

# M.Duke 
# UKOLN March 2003 
# Version 0.1

# Modifies RDN XML records

# Usage trans.pl -c <configfile> -i <inputfile> -o <outputfile>

# applies the regular expressions in configfile to the
# input file and writes the output to outputfile
# Note: inputfile must be given with the full path,
# including the current directory e.g. `./record.xml'
# if the -i switch is - the input is taken from standard input

# Reads the required transformations from a config file
# the config file is made up of element_name regexp pairs
# each pair is on a new line
# e.g. type s/xxx/yyyy/i 
# and writes the output to the output file
# deals with input files containing more than one record
# does not deal with multiple files

use  XML::Simple;
use Data::Dumper;
use Getopt::Std;

if (! getopts('c:i:o:')){
  die ("Usage: trans.pl -c configfile -i input file or - for standard input -o output file");
}
print "input file is $opt_i config file is $opt_c output file is $opt_o\n";

if ($opt_i ne '-') { # an input file name has been given
  $xmlp = XMLin($opt_i, forcearray=>1);
} else { #xml is in standard input
    while(<>){
      $xmlin .= $_;  # store in a string
    }
  $xmlp = XMLin($xmlin, forcearray=>1); # give the string to XMLin
}

$xmlt = doTransforms($xmlp); # call this for each file

# for dealing with multiple files (how to handle output?)

sub doTransforms {  # applies to transforms to one file

  my ($ref) = @_;  #content of one or more records from one file

  @records = @{$ref->{record}}; #an array of references to records

  open (CONF, "$opt_c") || die "can't open config file $opt_c";
  # for each line in CONF
  while (<CONF>){
    chop;
    ($tag, $regexp) = split(/ /, $_);
print "Tag is $tag regexp is $regexp";
    #deal with records one by one
    foreach $rec (@records){ # could be each $r in $ref->record[$r]
      #print "First $tag found is $rec->{metadata}[0]->{dc}[0]->{$tag}->[0]\n";
      # do the substitutions only for matching tag 
      if ($rec->{metadata}[0]->{dc}[0]->{$tag}) {
        #print "found a $tag\n";
        # dereference and get array of the values of element
        @val = @{ $rec->{metadata}[0]->{dc}[0]->{$tag} };
        $i = 0;
        while ($i <= $#val){
	      #print "i is $i\n";
          print "Apply $regexp to $tag $rec->{metadata}[0]->{dc}[0]->{$tag}->[$i]\n";
#          $rec->{metadata}[0]->{dc}[0]->{$tag}->[$i] = trim($rec->{metadata}[0]->{dc}[0]->{$tag}->[$i]);
          $substituted = "\$rec->{metadata}[0]->{dc}[0]->{$tag}->[$i] =~ $regexp;";
          eval ($substituted);
	      print "After applying regexp content of $tag is $rec->{metadata}[0]->{dc}[0]->{$tag}->[$i]\n";

          $i++;
        } #finished substituting all values of a tag
      } #finished dealing with tag if found
    } #end of a record
  } # finished with config file
return $ref;
1
} #end of do Transform 


sub trim {
  ($_) = @_;
  s/^\s*//;
  s/\s*$//;
  return $_;
}

# finally print out the substituted XML to file

$str = XMLout($xmlt);
open(OUTFILE, ">$opt_o") || die ("can't open output file $opt_o");
print OUTFILE "$str";
#print Dumper $xmlt;