18 |
# - uzvici |
# - uzvici |
19 |
# |
# |
20 |
|
|
21 |
|
use strict; |
22 |
|
|
23 |
sub kgh { |
sub kgh { |
24 |
my ($pre,$replace,$post) = @_; |
my ($pre,$replace,$post) = @_; |
25 |
$replace =~ s/[cè]/k/g; |
$replace =~ s/[cè]/k/g; |
33 |
# suglasnici |
# suglasnici |
34 |
my $su = '[^aeiou]'; |
my $su = '[^aeiou]'; |
35 |
|
|
36 |
|
my $palatal = '(lj|nj|j|æ|¾|¹|¾|¹t|¾d)'; |
37 |
|
|
38 |
|
# glagolni razredi |
39 |
|
my $g_1r = '[td]'; |
40 |
|
my $g_2r = '[sz]'; |
41 |
|
my $g_3r = '[pb]'; |
42 |
|
my $g_4r = '[kgh]'; |
43 |
|
my $g_5r = '[nm]'; |
44 |
|
my $g_6r = '(:?nu|n)'; |
45 |
|
|
46 |
my %rules; |
my %rules; |
47 |
my %stem_words; |
my %stem_words; |
48 |
my $words = 0; |
my $words = 0; |
49 |
my $stems = 0; |
my $stems = 0; |
50 |
|
|
51 |
|
|
52 |
|
my $last_stem = ''; |
53 |
|
my $errors = 0; |
54 |
|
sub check_stem { |
55 |
|
my $s = shift || return; |
56 |
|
if ($last_stem) { |
57 |
|
if ($last_stem ne $s) { |
58 |
|
print "ERROR==> "; |
59 |
|
$errors++; |
60 |
|
} |
61 |
|
} else { |
62 |
|
$last_stem = $s; |
63 |
|
} |
64 |
|
} |
65 |
|
|
66 |
while(<>) { |
while(<>) { |
67 |
chomp; |
chomp; |
68 |
|
next if (/^#/); |
69 |
if (/^$/) { |
if (/^$/) { |
70 |
print "\n"; |
print "\n"; |
71 |
|
$last_stem = ''; |
72 |
next; |
next; |
73 |
} |
} |
74 |
|
|
76 |
|
|
77 |
my $orig = $_; |
my $orig = $_; |
78 |
|
|
|
# imenice |
|
|
# vrsta a |
|
79 |
|
|
80 |
unless ( |
unless ( |
81 |
|
# infinitiv |
82 |
|
s/(${sa})(ti|æi)$/$1.$2 100/g || |
83 |
|
# 1. razred |
84 |
|
s/([^s])[td](em|e¹|emo|ete|oh|osmo|oste|o¹e|ijah|ija¹e|ijasno|ijaste|ijahu|imo|ite|en|ena|eni)$/$1.sti 101/g || |
85 |
|
# 2. razred |
86 |
|
s/(${sa}[sz])(em|e¹|e¹emo|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)$/$1.sti 102/ || |
87 |
|
# 3. razdred |
88 |
|
s/(p|b|sp|su)(em|e¹|e|emo|ete|u|oh|osmo|oste|o¹e|ah|a¹e|asmo|aste|ahu|ijah|ija¹e|i|imo|ite||uæi|av¹i|ao|la|lo|en|ena|eni)$/$1.sti 103/ || |
89 |
|
|
90 |
|
# imenice |
91 |
|
|
92 |
|
# vrsta a |
93 |
|
s/(${su}st)a$/$1 13/g || |
94 |
|
s/(${su})c[ae]$/$1ce 17/g || |
95 |
# kgh |
# kgh |
96 |
s/(${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,' 1')/gex || |
s/(${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,' 1')/gex || |
97 |
s/(${sa}[kgh])(a|u|om)$/$1 2/g || |
s/(${sa}[kgh])(a|u|om)$/$1 2/g || |
98 |
s/(${su})([è¾¹czs])(i|e|ima)$/kgh($1,"a$2",' 3')/gex || |
s/(${su})([è¾¹czs])(i|e|ima)$/kgh($1,"a$2",' 3')/gex || |
99 |
|
# imenice na palatal |
100 |
|
s/${palatal}${palatal}(a|u|em|i|ima|e)$/$1a$2 9/g || |
101 |
|
s/${palatal}(a|u|em)$/$1 8/g || |
102 |
# nepostojano a |
# nepostojano a |
103 |
s/(${su}a{$su})a/$1 4/g || |
s/(${su}a{$su})a/$1 4/g || |
104 |
s/(${su})(${su})(a|u|i|e|om|ima)$/$1a$2 5/g |
s/(${su})(${su})(a|u|i|e|om|ima)$/$1a$2 5/g |
105 |
) { |
) { |
106 |
|
|
107 |
|
# vrsta a |
108 |
|
s/me$/me 11/g || |
109 |
|
s/(eta|ena)$/e 12/g || |
110 |
|
s/(${sa}${su})(o|e|a|u|om|em|i|ima|ina)$/$1 7/g || |
111 |
|
|
112 |
|
s/(${su})sa$/$1as 14/g || |
113 |
|
s/(${su})ena$/$1e 16/g || |
114 |
|
s/(${su})eta$/$1e 17/g || |
115 |
|
s/(${su})([oe])$/$1-$2 18/g || |
116 |
|
|
117 |
# mno¾ina |
# mno¾ina |
118 |
s/(${su})(ov|ev)*(i|a|ima|e|in|ina)$/$1 6/g; |
s/(${su})(ov|ev)*(i|a|ima|e|in|ina|eta)$/$1 6/g; |
|
# jednina |
|
|
s/(o|e|a|u|om|em)$/ 7/g; |
|
119 |
} |
} |
120 |
|
|
121 |
if (s/^(.+)\s(\d+)$/$1\t$2/g) { |
if (s/^(.+)\s(\d+)$/$1\t$2/g) { |
122 |
$rules{$2}++; |
$rules{$2}++; |
123 |
$stems++; |
$stems++; |
124 |
$stem_words{$1}++; |
$stem_words{$1}++; |
125 |
|
check_stem($1); |
126 |
|
} else { |
127 |
|
$last_stem = $_; |
128 |
} |
} |
129 |
|
|
130 |
printf("%-15s %s\n",$orig,$_); |
printf("%-15s %s\n",$orig,$_); |
131 |
|
|
132 |
} |
} |
133 |
my $nr_stems = keys(%stem_words); |
my $nr_stems = keys(%stem_words); |
134 |
printf "\n# %d words, %d stems in %d ops, %.2f%% size\n",$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; |
135 |
|
|
136 |
|
foreach my $s (keys %stem_words) { |
137 |
|
print "#stem $stem_words{$s} $s\n"; |
138 |
|
} |
139 |
|
|
140 |
|
foreach my $r (sort keys %rules) { |
141 |
|
print "#rule $rules{$r} $r\n"; |
142 |
|
} |