#!/usr/bin/perl -w
###########################################################################
#
# Corpus Analysis and Programming, Homework #6.
# Conversion from numeric values to corresponding symbolic names.
#
# Data source for this program is adapted from the following:
#   1. CKIP tagset: http://www.sinica.edu.tw/~tdbproj/kiwi/use05.html
#   2. testing text: file "ev1" in Academia Sinica balanced corpus, v3.0.
#
# Programmed by William Yeh, 2002/10/16.
#
###########################################################################

#
# global definitions
#

$Tagset_Filename = "ckip-tagset.txt";  # the CKIP tagset
# @Tag_Mapping;                        # number -> tag mapping


#
# first, check for command arguments
#

if (scalar(@ARGV) != 2) {
  die "Usage: s3_tag.pl  [input filename]  [output filename]";
}


#
# second, read in table of SKIP tagset
#

open(TAGSET, "$Tagset_Filename")
  or die "Fail to open $Tagset_Filename: $!";

while ($line = <TAGSET>) {
  chomp $line;

  next if $line =~ /^\s*$/;  # skip blank lines
  next if $line =~ /^\s*#/;  # skip comment lines

  $line =~ /^(\d+)\s+([^\s]+)/;
  ($number, $tag) = ($1, $2);
  # print "$number\t$tag\n";

  $Tag_Mapping[ $number ] = $tag;
}

close TAGSET;


#
# third, process input text...
#

open(INPUT,  $ARGV[0])      or die "Fail to open $ARGV[0]: $!";
open(OUTPUT, "> $ARGV[1]")  or die "Fail to open $ARGV[1]: $!";


while ($line = <INPUT>) {
  chomp $line;

  if ($line =~ /^\s*$/) {  # blank lines: print as-is
      print OUTPUT "\n";
      next;
  }

  $line =~ /^([^\s]+)\s+([\d]+)/;
  ($word, $number) = ($1, $2);

  print OUTPUT  $word, "\t", $Tag_Mapping[ $number ], "\n";
}

close INPUT;
close OUTPUT;