#! /usr/bin/perl  -w
# Light stemmer for the Russian language (to be viewed by selecting UTF-8 encoding)
# It removes case endidngs from nouns and adjectives
#
# We assume that each character (in Cyrillic) needs two bytes
#
# done by Lj. Dolamic University of Neuchatel (www.unine.ch/info/clef/)

while($line = <>){
	chomp $line;
	$line =~ s/^()+//;
	$stem = RussianLightStemming($line);
	if(length($line) > 0){
		print "$line  --> $stem \n";
	}
	else{
		print "\n";
	}
	
}
exit(0);

sub RussianLightStemming {
	my($line);
	$line = $_[0];
	
	if(length($line) < 8){ # do  nothing with words having less then 4 characters
		return($line);		
	}
	$line = remove_case($line);
	
	#normalization
	
	$line =~ s/нн$/н/;
	$line =~ s/ь$//;	
	return($line);
}

sub remove_case(){
	my($word,$i);
	$word =$_[0];
	$i = length($word);
	
	
	if($word =~ m/иями$/){
		return(substr($word,0,$i-8))
		
	}
	if($word =~ m/оями$/){
		return(substr($word,0,$i-8))
		
	}
	if($word =~ m/оиев$/){
		return(substr($word,0,$i-8))
		
	}
	if($word =~ m/иях$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/иям$/){
		return(substr($word,0,$i-6))
		
	}
	
	if($word =~ m/ями$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/ами$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, G
	
	if($word =~ m/его$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, D
	if($word =~ m/ему$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/ери$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard stem plural I
	if($word =~ m/ими$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/иев$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard  stem musculine and neuter, G
	if($word =~ m/ого$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard stem musculine and neuter, D
	if($word =~ m/ому$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# hard stem plural, I
	if($word =~ m/ыми$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/оев$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/оям$/){
		return(substr($word,0,$i-6))
		
	}
	if($word =~ m/оях$/){
		return(substr($word,0,$i-6))
		
	}
	# adjectives
	# soft stem feminine, N
	if($word =~ m/яя$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ях$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# soft stem feminine,A
	if($word =~ m/юю$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem feminine,N
	if($word =~ m/ая$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ах$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem feminine,I
	if($word =~ m/ею$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem plural, G, L
	if($word =~ m/их$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ия$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem feminine, I
	if($word =~ m/ию$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ою$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem feminine, A
	if($word =~ m/ую$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ям$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ых$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ея$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ам$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem neuter, N, G
	if($word =~ m/ее$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem feminine, G, D, L, I
	if($word =~ m/ей$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, L
	if($word =~ m/ем$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ём$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ев$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ие$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem musculine, N, G
	if($word =~ m/ий$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, I
	# hard stem plural, D
	if($word =~ m/им$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem neuter, N, G
	if($word =~ m/ое$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem musculine , N
	# hard stem feminine , G, D, L, I
	if($word =~ m/ой$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, L
	if($word =~ m/ом$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ов$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem plural, G, L
	if($word =~ m/ые$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard stem musculine, N
	if($word =~ m/ый$/){
		return(substr($word,0,$i-4))
		
	}
	# adjectives
	# hard and soft stem musculine and neuter, I
	# hard stem plural, D
	if($word =~ m/ым$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ми$/){
		return(substr($word,0,$i-4))
		
	}
	if($word =~ m/ю$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/й$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/ы$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/я$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/а$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/е$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/и$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/о$/){
		return(substr($word,0,$i-2))
		
	}
	if($word =~ m/у$/){
		return(substr($word,0,$i-2))
		
	}
	
	
}