#############################################################
# Program metapro.pl
#
# Grabs values from metadata records and builds counts for
# each value
#
# Author: Archie Warnock (warnock@awcubed.com) May 2004
#############################################################

use strict;
my %key_counts;
my $xml;
my $key;
my $file;

# Get the list of all matching files
use Cwd;
my $current_dir = &cwd();
my $datadir= $ARGV[0];		 		 # Command-line argument
chdir $datadir;		 		 		 # change to the input directory
my @filelist = <*.xml>;		 		 # get the list of all XML files there
undef $/;		 		 		 # set up to slurp in the entire file

if ( $datadir eq "" ) {
    print "\nUsage: metapro.pl \<directory_name_with_FGDC_XML\>
Writes to screen unless redirected to a file.";
    exit;
}

foreach $file (@filelist) {
    open (XML, $file);		 		 # open one file at a time
    $xml = <XML>;		 		 # read in the XML as one big string

    # Grab the contents of the <themekey> element
    ($key) = ($xml =~ m#<themekey>\s*(.*?)\s*</themekey>#is);
    
    # Initialize the hash value, if not already defined
    unless (defined($key_counts{$key})) {
		 $key_counts{$key} = 0;
    }

    $key_counts{$key}++;        # Now increment the count
    close XML;		                 # and clean up by closing the file
}

chdir $current_dir;		 		 # switch back to the original directory

# Print out the results.  This is tricky - it uses the numeric
# comparison function to compare the frequencies and sorts the
# keys on the results.  As a result, the loop over the hash grabs
# the keys in the order of the frequencies.
my $sortkeys;
foreach $sortkeys 
    (sort {$key_counts{$b} <=> $key_counts{$a} } 
     keys %key_counts) {
		 print "$sortkeys ($key_counts{$sortkeys}).\n";
}

print "---\n"