--- stem-hr.pl 2005/02/26 16:19:22 6 +++ stem-hr.pl 2005/02/26 19:06:04 10 @@ -35,6 +35,14 @@ 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; @@ -42,10 +50,14 @@ my $last_stem = ''; +my $errors = 0; sub check_stem { my $s = shift || return; if ($last_stem) { - print "ERROR==> " if ($last_stem ne $s); + if ($last_stem ne $s) { + print "ERROR==> "; + $errors++; + } } else { $last_stem = $s; } @@ -64,14 +76,24 @@ my $orig = $_; - # imenice unless ( + # infinitiv + s/(\w)(ti|æi)$/$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)$/$1s.ti 102/ || + # 3. razdred + s/(p|b|sp|su)(em|e¹|e|emo|ete|u|oh|osmo|oste|o¹e|ah|a¹e|asmo|aste|ahu|ijah|ija¹e|ijahu|i|imo|ite||uæi|av¹i|ao|la|lo|en|ena|eni)$/$1s.ti 103/ || + + # imenice + # vrsta a s/(${su}st)a$/$1 13/g || s/(${su})c[ae]$/$1ce 17/g || # kgh - s/(${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,' 1')/gex || + s/(\w${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$2",' 3')/gex || # imenice na palatal @@ -85,15 +107,16 @@ # vrsta a s/me$/me 11/g || s/(eta|ena)$/e 12/g || - s/(${sa}${su})(o|e|a|u|om|em|i|ima|ina)$/$1 7/g || + + s/(\w${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-$2 18/g || + s/(${su})([oe])$/$1$2 18/g || + + 0; - # mno¾ina - s/(${su})(ov|ev)*(i|a|ima|e|in|ina|eta)$/$1 6/g; } if (s/^(.+)\s(\d+)$/$1\t$2/g) { @@ -109,7 +132,7 @@ } my $nr_stems = keys(%stem_words); -printf "\n# %d words, %d stems in %d ops, %.2f%% size\n",$words,$nr_stems,$stems,($nr_stems*100/$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";