#!/local/bin/perl -w
# Split a file or stream comming in, saving each part in a separate file.
# Command line parameters:  output_directory: where to save the output
#                           pattern: regular expression for the file separator
#                                    (must match the whole separator line)
# Saves each part using the three first characters of the separator.
# This is intended to split out a Brown corpus that has been merged with a
# single line marking the name of the file at the beginning of each file
#
# Note:  any text coming before the first file separator is dumped to
#        STDERR

#use strict "vars";
use strict;

sub PrintUsageAndExit {
    print STDERR
        "Usage: $0 output_dir file_separator\n";
    exit;
}

if ( $#ARGV != 1 ) { PrintUsageAndExit(); }
my $outdir = $ARGV[0];
my $separator = $ARGV[1]; 
shift; shift;

select STDERR; $| = 1;

print STDERR "outdir $outdir\nseparator $separator\n";

open OUT, ">&STDERR";

while (<>) {
    if ( /\A($separator)\s*\Z/ ) {
        close OUT;
        my $filename = "$outdir/" . substr($1, 0, 3);
        open OUT, ">$filename.tag" or die "Can't open $filename: $!\n";
        next;
    }

    print OUT;

}