#!/usr/bin/perl

use warnings;
use strict;

die "Usage: $0 <from dir> <to dir>\n" if @ARGV != 2;

# ==================
# my system function
# ==================

sub mysystem {
    my $ret = system @_;

    die "Interrupted\n" if $ret;
}

# =====================
# function definitions
# =====================

sub reduce {
    my $fname = shift;
    my $tmp   = $fname.'.tmp';

    open FILE1, "<$fname";
    open FILE2, ">$tmp";

    <FILE1>; # strip off the first line

    while (<FILE1>) {
	my @l = split /\s+/;

	print FILE2 "$l[2]\n";
    }

    close FILE2;
    close FILE1;

    mysystem qq{mv $tmp $fname};
}

sub convert {
    my ($fromdir, $todir) = @_;

    warn "Copying directory structure from $fromdir to $todir\n";

    opendir DIR, $fromdir;
    my @alldirs = grep { !/^\.+/ } (readdir DIR);
    closedir DIR;
    
    mysystem qq{mkdir -p $todir};
    mysystem qq{mkdir -p $todir/$_} foreach (@alldirs);
    
    foreach my $dir (@alldirs) {
	my $from = "$fromdir/$dir";
	my $to   = "$todir/$dir";
	
	opendir DIR, $from;
	
	while ($_ = readdir DIR) {
	    if (/^([1-9oz][ab])\.wav$/) {
		my $f1 = "$from/$1.wav";
		my $f2 = "$to/$1.dat";

		mysystem qq{sox $f1 -r 8000 $f2 polyphase};

		&reduce($f2);
	    }
	}
	
	closedir DIR;
    }
}

# ================
# main directories
# ================

my $fromdir = shift;
my $todir   = shift;

my $datadir   = "$todir/data";
my $listdir   = "$todir/lists";

# =====================================
# find original train and test datasets
# =====================================

my @l = split /\n/, qx{find $fromdir -name man};

my ($traindir) = grep {/train/} @l;
my ($testdir)  = grep {/test/} @l;

# ==============================
# create train and test datasets
# ==============================

&convert($traindir, "$datadir/clean/train/man");
&convert($testdir, "$datadir/clean/test/man");

# ================
# build list files
# ================

my $tmp1 = "$listdir/clean/train";
my $tmp2 = "$listdir/clean/test";

mysystem qq{mkdir -p $tmp1 $tmp2};

foreach my $id (1..9, 'o', 'z') {
    mysystem qq{find $datadir/clean/train -name $id?.dat >$tmp1/$id.lst};
    mysystem qq{find $datadir/clean/test -name $id?.dat >$tmp2/$id.lst};
}