countpypos.pl

采用复杂数据结构从文件中生成一个带有拼音和词性信息的词典

open(In,"3000TagPY.txt");
$Num=1;
while(<In>){
	chomp;
	if ( $Num%3 == 2 ){
		@Word=/(\S+)\([^)\/]+\)\/\S+/g;
		@PY=/\S+\(([^)\/]+)\)\/\S+/g;
		@POS=/\S+\([^)\/]+\)\/(\S+)/g;
		for ($i=0;$i<@Word;$i++){
			${${$Hash{$Word[$i]}}[0]}{$POS[$i]}++;
			${$Hash{$Word[$i]}->[1]}{$PY[$i]}++;
		}
	}
	$Num++;	
}
close(In);

foreach (sort keys %Hash){
		$Ref=$Hash{$_};
		print "\#$_\n";
		foreach (sort keys %{$$Ref[0]} ){
			print "$_\t";
		}
		print "\n";

		foreach (sort keys %{$$Ref[1]} ){
			print "$_\t";
		}
		print "\n";
		
}