--- stem-hr.pl	2005/02/25 23:14:44	1
+++ stem-hr.pl	2005/07/09 10:04:51	11
@@ -18,16 +18,14 @@
 # - uzvici
 #
 
-sub rm_a {
-	s/a//g;
-}
+use strict;
 
 sub kgh {
 	my ($pre,$replace,$post) = @_;
 	$replace =~ s/[cè]/k/g;
 	$replace =~ s/[z¾]/g/g;
 	$replace =~ s/[s¹]/h/g;
-	return $pre . $replace . $post . "\tkgh";
+	return $pre . $replace . $post;
 }
 
 # samoglasnici
@@ -35,30 +33,116 @@
 # suglasnici
 my $su = '[^aeiou]';
 
+my $palatal = '(lj|nj|j|æ|¾|¹|¾|¹t|¾d)';
+
+# glagolni razredi
+my $g_1r = '[td]';
+my $g_2r = '[sz]';
+my $g_3r = '[pb]';
+my $g_4r = '[kgh]';
+my $g_5r = '[nm]';
+my $g_6r = '(:?nu|n)';
+
+my %rules;
+my %stem_words;
+my $words = 0;
+my $stems = 0;
+
+
+my $last_stem = '';
+my $errors = 0;
+sub check_stem {
+	my $s = shift || return;
+	if ($last_stem) {
+		if ($last_stem ne $s) {
+			print "ERROR==> ";
+			$errors++;
+		}
+	} else {
+		$last_stem = $s;
+	}
+}
+
 while(<>) {
 	chomp;
+	next if (/^#/);
+	if (/^$/) {
+		print "\n";
+		$last_stem = '';
+		next;
+	}
+
+	$words++;
 
 	my $orig = $_;
 
-	# imenice
-	# vrsta a
 
 	unless (
+		# infinitiv
+		s/(\w)(ti|æi)$/$1.$2 100/g ||
+		# 1. razred
+		s/([^sk])[td](em|e¹|e|emo|ete|oh|osmo|oste|o¹e|ijah|ija¹e|ijasno|ijaste|ijahu|imo|ite|en|ena|eni)$/$1s.ti 101/g ||
+		# 2. razred
+		s/(${sa})[sz](em|e¹|e|e¹emo|emo|ete|u|oh|e|osmo|oste|o¹e|ijah|ija¹e|ijasmo|ijaste|ijahu|imo|ite|uæi|av¹i|ao|la|lo|li|le|la|en|ena|eni)$/$1s.ti 102/ ||
+		# 3. razred
+		s/(p|b|sp|su)(em|e¹|e|emo|ete|u|oh|osmo|oste|o¹e|ah|a¹e|asmo|aste|ahu|ijah|ija¹e|ijahu|i|imo|ite||uæi|av¹i|ao|la|lo|en|ena|eni)$/$1s.ti 103/ ||
+		# 4. razred
+		s/[è¾¹](em|e¹|e|emo|ete|u|ah|ahu|en|ena)$/.æi 104/g ||
+		s/[k](oh|osmo|oste|o¹e|uæi|av¹i|ao|la|lo)$/.æi 105/g ||
+		s/[c](ijah|ija¹e|ijasmo|ijaste|ijahu|i|imo|ite)$/.æi 106/g ||
+		s/[g](nuti|oh|nuh|nu|av¹i|nuv¹i|ao|nuo|nem|ne¹|ni|imo|nut|nimo)$/.æi 107/g ||
+
+		# imenice
+
+		# vrsta a
+		s/(${su}st)a$/$1 13/g ||
+		s/(${su})c[ae]$/$1ce 17/g ||
+		# kgh
+		s/(\w${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,' 1')/gex ||
+		s/(${sa}[kgh])(a|u|om)$/$1 2/g ||
+		s/(${su})([è¾¹czs])(i|e|ima)$/kgh($1,"a$2",' 3')/gex ||
+		# imenice na palatal
+		s/${palatal}${palatal}(a|u|em|i|ima|e)$/$1a$2 9/g ||
+		s/${palatal}(a|u|em)$/$1 8/g ||
 		# nepostojano a
-		s/(${su}a{$su})a/$1 1/g ||
-		s/(${su})(${su})(a|u|i|e|om|ima)$/$1a$2\t3/g
+		s/(${su}a{$su})a/$1 4/g ||
+		s/(${su})(${su})(a|u|i|e|om|ima)$/$1a$2 5/g
 	) {
-		# kgh
-		s/(${su}${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,'')/gex;
-#		s/${su}${sa}[c](i|ima) 4//g;
-		s/(${su}${sa}[kgh])(a|u|om)\t5/$1/g;
-
-		# mno¾ina
-		s/(${su})(ov|ev)*(i|a|ima|e|in|ina)$/$1\t6/g;
-		# jednina
-		s/(o|e|a|u|om|em)$/\t7/g;
+
+		# vrsta a
+		s/me$/me 11/g ||
+		s/(eta|ena)$/e 12/g ||
+
+		s/(\w${sa}${su})(o|e|a|u|om|em|i|ima|ina|eta)$/$1 7/g ||
+
+		s/(${su})sa$/$1as 14/g ||
+		s/(${su})ena$/$1e 16/g ||
+		s/(${su})eta$/$1e 17/g ||
+		s/(${su})([oe])$/$1$2 18/g ||
+
+		0;
+
 	}
 
+	if (s/^(.+)\s(\d+)$/$1\t$2/g) {
+		$rules{$2}++;
+		$stems++;
+		$stem_words{$1}++;
+		check_stem($1);
+	} else  {
+		$last_stem = $_;
+	}
 
 	printf("%-15s %s\n",$orig,$_);
+
+}
+my $nr_stems = keys(%stem_words);
+printf "\n# %d words, %d stems in %d ops, %.2f%% size [%d errors]\n",$words,$nr_stems,$stems,($nr_stems*100/$words),$errors;
+
+foreach my $s (keys %stem_words) {
+	print "#stem $stem_words{$s} $s\n";
+}
+
+foreach my $r (sort keys %rules) {
+	print "#rule $rules{$r} $r\n";
 }