convertfileinput.pl

来自「一款数据挖掘的软件」· PL 代码 · 共 41 行

if ($#ARGV < 1) {
	print "Usage: ConvertFileInput.pl SRC OUTPUT\n";
	exit(1);
}

$SRC = $ARGV[0];
$OUTPUT = $ARGV[1];
#$MINFEATURENUM = $ARGV[2];
#$MAXFEATURENUM = $ARGV[3];

open SRC or die "Cannot open $SRC1 for reading\n";
open OUTPUT, ">$OUTPUT" or die "Cannot open $OUTPUT for writing\n";

$MAXFEATURENUM = 0;
$MINFEATURENUM = 1e7;
while ($LINE = <SRC>) {
   $LINE =~ tr/ /\:/;
   @sLine = split(/:/, $LINE);
   for ( $i = 1; $i < $#sLine; $i += 2) {
   	$MAXFEATURENUM = $sLine[$i] if ($MAXFEATURENUM < $sLine[$i]);
   	$MINFEATURENUM = $sLine[$i] if ($MINFEATURENUM > $sLine[$i]);
   }
}
close(SRC);
print "Min feature #: $MINFEATURENUM, Max feature #: $MAXFEATURENUM\n";

open SRC or die "Cannot open $SRC1 for reading\n";
while ($LINE = <SRC>) {
   chomp($LINE);	
   $LINE =~ tr/ /\:/;
   @sLine = split(/:/, $LINE);
   @sLine1 = ();
   for ( $i = 1; $i < $#sLine; $i += 2) {
   	$sLine1[$sLine[$i] - $MINFEATURENUM] = $sLine[$i+1]; 
   }
   $sLine1[$MAXFEATURENUM + 1 - $MINFEATURENUM] = $sLine[0];
   	
   print OUTPUT join(",", @sLine1), "\n";
}
close(SRC);