## Deals with  spelling variation & simple SANDHI for TELUGU --TEL-TAM MT##
#Oct, 2008, PARAMESHKRISHNA
#Then adopted for Telugu-Hindi by  gurao 1May2012

while($line=<>)
{
chomp($line);
	my $word1,$word2, $word;
	##my $word1, $word;
	$line=~s/\x{2018}/'/g; # <2018> ‘ is Replaced by single quote "'"
        $line=~s/\x{2019}/'/g; # <2019> ’ is Replaced by single quote "'"
        $line=~s/\x{200D}//g; # <200D> is Removed
        $line=~s/\x{200C}//g; # <200C> is Removed
        $line=~s/\x{200c}//g; # <200C> is Removed
        $line=~s/\x{202c}//g; # <200C> is Removed
        $line=~s/\x{200B}//g; # <200B> is Removed
        $line=~s/\x{feff}//g; # <feff> is Removed
        $line=~s/\x{0D}//g; # ^M is Removed
        $line=~s/á//g; # 

	$line=~/([0-9]*)\t(.*)/;
	$num=$1;
	$word=$2;

	if($word=~/([A-z])([vy]+)AgixeV\t/) {	##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
	$word=~/(.*)([vy]+)AgixeV(.*)/;
	$word=$1."\tunk";
	$word1="AgixeV".$3;
	}
        if($word=~/([A-z])([x]+)AgixeV\t/) {   ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)([x]+)AgixeV(.*)/;
        $word=$1."xa\tunk";
        $word1="AgixeV".$3;
        }
        if($word=~/([A-z])([vy]+)AgiveV\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)([vy]+)AgiveV(.*)/;
        $word=$1."\tunk";
        $word1="AgiveV".$3;        
        }
      #  if($word=~/([A-z])([vy]+)Agiruva\t/) {  
      #  $word=~/(.*)([vy]+)Agiruva(.*)/;
      #  $word=$1."\tunk";
      #  $word1="Agiruva".$3;
      #  }
        if($word=~/wweVMba\t/) {  ##mAdiwweVMba
        $word=~/(.*)wweVMba(.*)/;
        $word=$1."wwu\tunk";
        $word1="eVMba".$2;
        }
 	if($word=~/yukwa\t/) {  ##PAyibarayukwa
        $word=~/(.*)yukwa(.*)/;
        $word=$1."\tunk";
        $word1="yukwa".$2;
        }
        if($word=~/([nw]+)eVMba\t/) {  ##hoyiweVMba
        $word=~/(.*)([nw]+)eVMba(.*)/;
        $word=$1."u\tunk";
        $word1="eVMba".$2;
        }
        if($word=~/yenalla\t/) {  ##kammiyenalla
        $word=~/(.*)yenalla(.*)/;
        $word=$1."\tunk";
        $word1="alla".$2;
        }
 
        if($word=~/igiMwalU\t/) {  ##savAligiMwalU
        $word=~/(.*)igiMwalU(.*)/;
        $word=$1."\tunk";
        $word1="giMwalU".$2;
 }
       if($word=~/veVMba\t/) {  ##hoyiweVMba
        $word=~/(.*)veVMba(.*)/;
        $word=$1."\tunk";
        $word1="eVMba".$2;
        }
        if($word=~/wweVMxu\t/) {  ##mAdiwweVMxu
        $word=~/(.*)wweVMxu(.*)/;
        $word=$1."wwu\tunk";
        $word1="eVMxu".$2;
        }
        if($word=~/([n]+)eVMxu\t/) {  ##mAdiwweVMxu
        $word=~/(.*)eVMxu(.*)/;
        $word=$1."u\tunk";
        $word1="eVMxu".$2;
        }
        if($word=~/([yv]+)eVMxu\t/) {  ##mAdiwweVMxu
        $word=~/(.*)([yv]+)eVMxu(.*)/;
        $word=$1."\tunk";
        $word1="eVMxu".$2;
        }
        if($word=~/AreVMxu\t/) {  ##ittixxAreVMxu
        $word=~/(.*)AreVMxu(.*)/;
        $word=$1."AreV\tunk";
        $word1="eVMxu".$2;
        }
        if($word=~/iveVlla\t/) {  ##iveVlla
        $word=~/(.*)iveVlla(.*)/;
        $word=$1."ivu\tunk";
        $word1="eVlla".$2;
        }
        if($word=~/ংোxalleVlla\t/) {  ##iveVlla
        $word=~/(.*)xalleVlla(.*)/;
        $word=$1."xalli\tunk";
        $word1="eVlla".$2;
        }
        if($word=~/eVllara\t/) {  ##mAdiwweVMba
        $word=~/(.*)eVllara(.*)/;
        $word=$1."a\tunk";
        $word1="eVllara".$2;
        }
        if($word=~/oVmmeV\t/) {  ##yAvAgalAxaroVmmeV
        $word=~/(.*)oVmmeV(.*)/;
        $word=$1."U\tunk";
        $word1="oVmmeV".$2;
        }
        if($word=~/ixxavu\t/) {  ##maragagalYuixxavu
        $word=~/(.*)ixxavu(.*)/;
        $word=$1."u\tunk";
        $word1="ixxavu".$2;
        }
        if($word=~/([yv]+)illaxeV\t/) {  ##AyXavillaxeV
        $word=~/(.*)([yv]+)illaxeV(.*)/;
        $word=$1."\tunk";
        $word1="illaxeV".$3;
        }
        if($word=~/([yv]+)illa\t/) {  ##AyXavillaxeV
        $word=~/(.*)([yv]+)illa(.*)/;
        $word=$1."\tunk";
        $word1="illa".$3;
        }
        if($word=~/([tvy]+)ixeV\t/) {  ##AScaryakaravAxuxu modified to include AScaryakara([vAxuxu])emo yash
        $word=~/(.*)ixeV(.*)/;
        $word=$1."u\tunk";
        $word1="ixeV".$2;
        }
        if($word=~/allixeV\t/) {  
        $word=~/(.*)allixeV(.*)/;
        $word=$1."alli\tunk";
        $word1="ixeV".$2;
        }
        if($word=~/galYiveV\t/) {  ##AScaryakaravAxuxu modified to include AScaryakara([vAxuxu])emo yash
        $word=~/(.*)galYiveV(.*)/;
        $word=$1."galYu\tunk";
        $word1="iveV".$2;
        }
        if($word=~/vAguwwaxeV\t/) {  ##AScaryakaravAxuxu modified to include AScaryakara([vAxuxu])emo yash
        $word=~/(.*)vAguwwaxeV(.*)/;
        $word=$1."\tunk";
        $word1="AguwwaxeV".$2;
        }
#	if($word=~/lAguwwaveV\t/) {  ##AScaryakaravAxuxu modified to include AScaryakara([vAxuxu])emo yash
 #       $word=~/(.*)lAguwwaveV(.*)/;
  #      $word=$1."lu\tunk";
   #     $word1="AguwwaveV".$2;
#	}
        if($word=~/yAguvuxilla\t/) {  ##AScaryakaravAxuxu modified to include AScaryakara([vAxuxu])emo yash
        $word=~/(.*)yAguvuxilla(.*)/;
        $word=$1."\tunk";
        $word1="Aguvuxilla".$2;
        }
        if($word=~/wwiruwwAreV\t/) {  ##baruwwiruwwAreV modified to include baru+wwA([iru+uww])emo yash & shanta 07012013
        $word=~/(.*)wwiruwwAreV(.*)/;
        $word=$1."wwA\tunk";
        $word1="iruwwAreV".$2;
        }
        if($word=~/vAgiruwwaxeV\t/) {  
        $word=~/(.*)vAgiruwwaxeV(.*)/;
        $word=$1."\tunk";
        $word1="AgiruwwaxeV".$2;
        }
      #  if($word=~/vAxareV\t/) {  ##ulleKavAxareV modified to include ulleka+v([AxareV])emo yash 29082013
      #  $word=~/(.*)vAxareV(.*)/;
      #  $word=$1."\tunk";
      #  $word1="AxareV".$2;
      #  }
      #  if($word=~/Ayiwu\t/) {  
      #  $word=~/(.*)Ayiwu(.*)/;
      #  $word=$1."a\tunk";
      #  $word1="Ayiwu".$2;
      #  }
        if($word=~/xallAyiwu\t/) {
        $word=~/(.*)xallAyiwu(.*)/;
        $word=$1."xalli\tunk";
        $word1="Ayiwu".$2;
	}
        if($word=~/allaxeVyU\t/) {
        $word=~/(.*)allaxeVyU(.*)/;
        $word=$1."u\tunk";
        $word1="allaxeV".$2;
        }

        if($word=~/alAyiwu\t/) {
        $word=~/(.*)alAyiwu(.*)/;
        $word=$1."u\tunk";
        $word1="Ayiwu".$2;
        }
        if($word=~/raMxu\t/) {       
        $word=~/(.*)raMxu(.*)/;
        $word=$1."\tunk";
        $word1="raMxu".$2;
        }
        if($word=~/াyara\t/) {
        $word=~/(.*)yara(.*)/;
        $word=$1."\tunk";
        $word1="raMxu".$2;
        }
        if($word=~/([A-z])([nryx]+)oVMxigeV\t/) {  ##GoRisixxAreV modified to include GoRisu([ixxAreV])emo yash & shanta 08012013
        $word=~/(.*)oVMxigeV(.*)/;
        $word=$1."a\tunk";
        $word1="oVMxigeV".$2;
        }
        if($word=~/vAgi\t/) {  ##GoRisixxAreV modified to include GoRisu([ixxAreV])emo yash & shanta 08012013
        $word=~/(.*)vAgi(.*)/;
        $word=$1."\tunk";
        }
        if($word=~/ilYixu\t/) {  ##keVlYagilYixu modified to include keVlYag(eV)ilYixu emo yash & shanta 08012013
        $word=~/(.*)ilYixu(.*)/;
        $word=$1."eV\tunk";
        $word1="ilYixu".$2;
        } 
        if($word=~/virisikoVlYlYabahuxu\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)virisikoVlYlYabahuxu(.*)/;
        $word=$1."\tunk";
        $word1="irisikoVlYlYabahuxu".$2;
        }
        if($word=~/iruwwaxeV\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)iruwwaxeV(.*)/;
        $word=$1."u\tunk";
        $word1="iruwwaxeV".$2;
        }
        if($word=~/iruwwareV\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)iruwwareV(.*)/;
        $word=$1."i\tunk";
        $word1="iruwwareV".$2;
        }
        if($word=~/iruvuxu\t/) {  
        $word=~/(.*)iruvuxu(.*)/;
        $word=$1."i\tunk";
        $word1="iruvuxu".$2;
        }
        if($word=~/xalliruva\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)xalliruva(.*)/;
        $word=$1."xalli\tunk";
        $word1="iruva".$2;
        }
        if($word=~/galYAgixxAreV\t/) {  ##marcipowAremo modified to include vaswA([dmnvr])emo gurao30052012
        $word=~/(.*)galYAgixxAreV(.*)/;
        $word=$1."galYu\tunk";
        $word1="AgixxAreV".$2;
        }
        if($word=~/([A-z])([vy]+)iruva\t/) {  ##GoRisixxAreV modified to include GoRisu([ixxAreV])emo yash & shanta 08012013
        $word=~/(.*)([vy]+)iruva(.*)/;
        $word=$1."\tunk";
        $word1="iruva".$3;
        }
        if($word=~/koVMdiruva\t/) {  ##GoRisixxAreV modified to include kattikoVMdu([iruva])emo yash
        $word=~/(.*)koVMdiruva(.*)/;
        $word=$1."koVMdu\tunk";
        $word1="iruva".$2;
        }
        if($word=~/([A-z])([y]+)oVlYageV\t/) {  ##GoRisixxAreV modified to include GoRisu([ixxAreV])emo yash & shanta 08012013
        $word=~/(.*)([y]+)oVlYageV(.*)/;
        $word=$1."\tunk";
        $word1="oVlYageV".$3;
        }

       # if($word=~/vallaxeV\t/) {  ##mAwravallaxeV modified to mAwra allaxeV emo yash & shanta 11022013
       # $word=~/(.*)vallaxeV(.*)/;
       # $word=$1."\tunk";
       # $word1="allaxeV".$2;
       # }
####################################################

	print "1\t$word\n";
	if($word1 ne "") {
	print "1\t$word1\n";
	}
	if(($word1 ne "")&&($word2 ne "")) {
	print "1\t$word2\n"; 
	$word2=~s/.*//g; }
}