use strict;
use XML::Simple;
use Data::Dumper;

#-------------#
#  Variables 
#-------------#

my $logfile="c:\\FTP_Scripts\\logfile.txt";
my $numArgs= $#ARGV + 1;
my $argnum;
my $error_time_stamp;
my $err;
my $debug=0;

my $xmlfile;   #name of xml file including file path
my $xmlfh;     #xml file handle 
my $csvfile;   #name of csv output file
my $csvfh;     #csv file handle
my $csvdir;    #csv file directory

my $current_line;

my @data_set;     # array to hold all the records in the xml file
my $dataset_id = 'REC id';  #sub-string used to identify the lines with data
my $current_timestamp;
my $previous_timestamp;




#-------------#
#  Main 
#-------------#

## Get command line parameters
if($numArgs < 1){
  print_usage();
  exit;
}

if($ARGV[0]){
  $xmlfile=$ARGV[0];
  $error_time_stamp = geterror_time_stamp();
  logit("$error_time_stamp > Input XML file: $xmlfile");
}

if($ARGV[1]){
  $csvdir=$ARGV[1];
  $error_time_stamp = geterror_time_stamp();
  logit("$error_time_stamp > Output directory: $csvdir");
}


## Open data file for reading
if( open($xmlfh, '<', "$xmlfile") ){
  $error_time_stamp = geterror_time_stamp();
  logit("$error_time_stamp >Successfully opened file: $xmlfile");
}
else{
  $error_time_stamp = geterror_time_stamp();
  logit("$error_time_stamp > Failed to open file: $xmlfile");
}

my $result;
my $line_num = 0;
## Get the lines with data, string starts with "^REC id.*"
while( $current_line = <$xmlfh> ){
  chomp ($current_line);
  # Identify the lines that begin with "REC id"
  $result = index($current_line,$dataset_id);
  if( $result == 1 ){
    $data_set[$line_num] = $current_line;
    $line_num = $line_num + 1;
  }
}

my $xml_ref;
my $xs;
my $x;
for($x==0; $x < $line_num; $x++){
  print "\n\n$x : $data_set[$x]\n";
  $xs = new XML::Simple->new();
  $xml_ref = $xs->XMLin($data_set[$x]);
  print Data::Dumper->Dump($xml_ref);
}

close $xmlfh;
exit 0;

#--------------#
# Sub-routines
#--------------#
sub logit {
  my $line=shift;
  open(LOG,">>$logfile") or die("$! Can't open $logfile");
  if($debug){
    print "$line\n";
  }
  print LOG "$line\n";
  close LOG;
}


sub geterror_time_stamp {
  my ($sec, $min, $hour, $day, $month, $yr19, @rest) = localtime(time);
  $error_time_stamp = $month . "-" . $day . "-" . ($yr19 + 1900) . " " . $hour . ":" . $min . ":" . $sec;
  return $error_time_stamp;
}

sub print_usage{
  print "\nNo files were specified,\nCorrect usage is:\n";
  print "xml-csv.pl  <XML input file>  [output directory, optional]\n\n";
  return;
}