#!/usr/local/bin/perl

# This script will encode a given list of files of the RM database
# store the results in a directory tree given the root
# It has options for coding training data and test data for both
# the speaker dependent and speaker independent training sets
#
# usage: coderm.prl ndxdir cdmount dbtype dbportion targetroot config codelist
#
#  ndxdir       - directory to find the NIST filelists
#  cdmount      - where CD-ROM with both data files is mounted
#  dbtype       - ind | dep
#  dbportion    - train | dev_aug | traina | trainb | evalset | dms0_tst
#  targetroot   - root directory for coded files
#  config       - configuration file used to set coding parameters
#  codelist     - list of coded files
#
#  for speaker dependent - dbportion = traina (1st 6 speakers)
#                                    = trainb (2nd 6 speakers)
#  for speaker independent - dbportion = train (main 72 speakers)
#                          - dbportion = dev_aug (augmented 37 speakers)
#  In both cases evalset is one of {feb89, oct89, feb91, sep92}
#
# eg coderm.prl $RMLIB/ndx $CDRM1 ind train /home/azure/pcw/rmdata
#
# The script first identifies the appropriate filelist from ndxdir
# and then codes all the files listed transforming the filenames 
# so that they are unique e.g  bef0_3/st0013.wav -> bef0_3_st0013.mfc
#
# Copyright (c) Phil Woodland + Julian Odell, 1995
# Last Updated 17/8/95
#
# convert to perl by TL 7/1998
#
push @INC, "$ENV{RMSCRIPTS}";
require "global.pl";                    # include global variables

if ( $#ARGV != 6 ) {
	die "usage: coderm.prl ndxdir cdmount dbtype dbportion targetroot config codelist.\n";
}

$ndxdir = $ARGV[0];
if ( ! -d $ndxdir ) {
   die "NIST filelist directory $ndxdir not found.\n";
}

$rmcd = $ARGV[1];
if ( ! -d $rmcd ) {
   die "RM CD-ROM $rmcd not found.\n";
}

$dbtype = $ARGV[2];
if ( $dbtype != ind && $dbtype != dep ) {
   die "dbtype of [ind | dep] expected and not $dbtype.\n";
}

$dbportion = $ARGV[3];
foreach ($dbportion) {
	/traina/ and do { 
		if ($dbtype == dep) {
			$ndx = "$ndxdir\\6a_deptr.ndx";
		}
		else {
			die "Illegal dbportion $dbportion for dbtype $dbtype.\n";
		}
		last;
	}; 
	/trainb/ and do { 
		if ($dbtype == dep) {
			$ndx = "$ndxdir\\6b_deptr.ndx";
		}
		else {
			die "Illegal dbportion $dbportion for dbtype $dbtype.\n";
		}
		last;
	}; 
	/train/ and do { 
		if ($dbtype == ind) {
			$ndx = "$ndxdir\\72_indtr.ndx";
		}
		else {
			die "Illegal dbportion $dbportion for dbtype $dbtype.\n";
		}
		last;
	}; 
	/dev_aug/ and do {
		if ($dbtype == ind) {
			$ndx = "$ndxdir\\37_indtr.ndx";
		}
		else {
			die "Illegal dbportion $dbportion for dbtype $dbtype.\n";
		}
		last;
	}; 
	/feb89/ and do {
		$ndx = "$ndxdir\\4_${dbtype}tst.ndx";
		last;
	};
	/oct89/ and do {
		$ndx = "$ndxdir\\5_${dbtype}tst.ndx";
		last;
	};
	/feb91/ and do {
		$ndx = "$ndxdir\\6_${dbtype}tst.ndx";
		last;
	};
	/sep92/ and do {
		$ndx = "$ndxdir\\7_${dbtype}tst.ndx";
		last;
	};
	/dms0_tst/ and do {
		$ndx = "$ndxdir\\dms0_tst.ndx";
		last;
	};
   die "dbtype not one of (train | dev_aug | traina | trainb | feb89 | oct89 | feb91 | sep92 | dms0_tst).\n";
}

$tgtroot = $ARGV[4];
if ( ! -d $tgtroot ) {
   die "Target root directory $tgtroot doesnt exist.\n";
}

$config = $ARGV[5];
$codefilelist = $ARGV[6];

if ( ! -r $ndx ) {
    die "Cant read RM filelist $ndx.\n";
}

$codescript = "${codefilelist}.code";

unlink $codefilelist;
unlink $codescript;

print "Making file lists\n";
$count = 0;
open(NDXFILE,"<$ndx") or die "Can't open file $ndx\n";
open(CODESCRIPT,">$codescript") or die "Can't open file $codescript\n";
open(CODEFILELIST,">$codefilelist") or die "Can't open file $codefilelist\n";
while (<NDXFILE>) {
	if (/^;/) {
		next;
	}
	s:^/(.*)/(.*)([/_])(.*)/(.*)/(.*)\.wav:$rmcd\\$1\\$2$3$4\\$5\\$6.wav $tgtroot\\$2_$4\\$5\\$5_$6.mfc:;
	printf CODESCRIPT;
	printf CODEFILELIST "$tgtroot\\$2_$4\\$5\\$5_$6.mfc\n";
	$dirname = "$tgtroot\\$2_$4\\$5";
	$newdir = 1;
	foreach $dirlist (@dirlists) {
		if ( $dirlist eq $dirname ) {
			$newdir = 0;
			last;
		}
	}
       
	if ( $newdir == 1 ) {
		@dirlists = ( @dirlists, $dirname );
	}
	$count++;
}
close(NDXFILE);
close(CODESCRIPT);
close(CODEFILELIST);
@coredirs = ("dep_eval", "dep_trn", "dev_aug", "ind_eval", "ind_trn");
foreach $coredir (@coredirs) {  
  if ( ! -d "$RMDATA\\$coredir" ) {
	system("mkdir $RMDATA\\$coredir");
  }
}

foreach $dirlist (@dirlists) {
	if ( ! -d $dirlist ) {
		system("mkdir $dirlist");
	}
}

print "Coding data\n";
system("HCopy -A -T 1 -C $config -S $codescript");
unlink $codescript;

print "Coding complete - $count files processed\n";