--- stem-hr.pl 2005/02/26 18:54:49 9 +++ stem-hr.pm 2005/07/09 10:15:09 12 @@ -1,4 +1,5 @@ -#!/usr/bin/perl -w +package StemHr; + # # Croatian stemmer # @@ -19,6 +20,7 @@ # use strict; +use locale; sub kgh { my ($pre,$replace,$post) = @_; @@ -43,39 +45,7 @@ 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 = $_; - +sub stem { unless ( # infinitiv @@ -84,8 +54,13 @@ s/([^sk])[td](em|eš|e|emo|ete|oh|osmo|oste|oše|ijah|ijaše|ijasno|ijaste|ijahu|imo|ite|en|ena|eni)$/$1s.ti 101/g || # 2. razred s/(${sa})[sz](em|eš|e|ešemo|emo|ete|u|oh|e|osmo|oste|oše|ijah|ijaše|ijasmo|ijaste|ijahu|imo|ite|ući|avši|ao|la|lo|li|le|la|en|ena|eni)$/$1s.ti 102/ || - # 3. razdred - s/(p|b|sp|su)(em|eš|e|emo|ete|u|oh|osmo|oste|oše|ah|aše|asmo|aste|ahu|ijah|ijaše|i|imo|ite||ući|avši|ao|la|lo|en|ena|eni)$/$1s.ti 103/ || + # 3. razred + s/(p|b|sp|su)(em|eš|e|emo|ete|u|oh|osmo|oste|oše|ah|aše|asmo|aste|ahu|ijah|ijaše|ijahu|i|imo|ite||ući|avši|ao|la|lo|en|ena|eni)$/$1s.ti 103/ || + # 4. razred + s/[čžš](em|eš|e|emo|ete|u|ah|ahu|en|ena)$/.ći 104/g || + s/[k](oh|osmo|oste|oše|ući|avši|ao|la|lo)$/.ći 105/g || + s/[c](ijah|ijaše|ijasmo|ijaste|ijahu|i|imo|ite)$/.ći 106/g || + s/[g](nuti|oh|nuh|nu|avši|nuvši|ao|nuo|nem|neš|ni|imo|nut|nimo)$/.ći 107/g || # imenice @@ -119,25 +94,6 @@ } - if (s/^(.+)\s(\d+)$/$1\t$2/g) { - $rules{$2}++; - $stems++; - $stem_words{$1}++; - check_stem($1); - } else { - $last_stem = $_; - } - - printf("%-15s %s\n",$orig,$_); - + s/\s\d+$//; } -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"; -}