#!/usr/bin/gawk -f

BEGIN{
  attr=0;
  class="class";
  dataLine=0;
  first=1;
  names=".names";
  data=".data";
  cfg=".cfg";
  file="file";
  IGNORECASE=1;
}

/%/{next;}

first{
  names=file names;
  data=file data;
  cfg=file cfg;
  first=0;
}

/@attribute/{
  sub ( /@attribute * /, "" );
  gsub( "'", "" );
  sub ( / \{/, ": " );
  sub ( /\}/, "" );
  sub ( "real", ": continuous" );
  sub ( "number", ": continuous" );
  sub ( "integer", ": continuous" );
  sub ( " :", ": " );
  attrs[attr++]=$0;
  class=$0;
}

dataLine{
  gsub( "'", "" );
  print $0 >> data;
}

/@data/{
  sub( /.* /, "", class );
  print class >> names;
  for( curAttr=0; curAttr < attr - 1; curAttr++ )
    print attrs[curAttr] >> names;
  dataLine=1;
}

END{
  print "granularity:    2" >> cfg;
  print "maxNumber:      3" >> cfg;
  print "maxSize:      100" >> cfg;
  print "randomTrials:  50" >> cfg;
  print "futileTrials:   5" >> cfg;
  print "bestClass:    30%" >> cfg;
}