18 |
# - uzvici |
# - uzvici |
19 |
# |
# |
20 |
|
|
21 |
sub rm_a { |
use strict; |
|
s/a//g; |
|
|
} |
|
22 |
|
|
23 |
sub kgh { |
sub kgh { |
24 |
my ($pre,$replace,$post) = @_; |
my ($pre,$replace,$post) = @_; |
33 |
# suglasnici |
# suglasnici |
34 |
my $su = '[^aeiou]'; |
my $su = '[^aeiou]'; |
35 |
|
|
36 |
|
my %rules; |
37 |
|
my %stem_words; |
38 |
|
my $words = 0; |
39 |
|
my $stems = 0; |
40 |
|
|
41 |
while(<>) { |
while(<>) { |
42 |
chomp; |
chomp; |
43 |
|
if (/^$/) { |
44 |
|
print "\n"; |
45 |
|
next; |
46 |
|
} |
47 |
|
|
48 |
|
$words++; |
49 |
|
|
50 |
my $orig = $_; |
my $orig = $_; |
51 |
|
|
68 |
s/(o|e|a|u|om|em)$/ 7/g; |
s/(o|e|a|u|om|em)$/ 7/g; |
69 |
} |
} |
70 |
|
|
71 |
s/\s(\d+)$/\t$1/g; |
if (s/^(.+)\s(\d+)$/$1\t$2/g) { |
72 |
|
$rules{$2}++; |
73 |
|
$stems++; |
74 |
|
$stem_words{$1}++; |
75 |
|
} |
76 |
|
|
77 |
printf("%-15s %s\n",$orig,$_); |
printf("%-15s %s\n",$orig,$_); |
78 |
|
|
79 |
|
} |
80 |
|
my $nr_stems = keys(%stem_words); |
81 |
|
printf "\n# %d words, %d stems in %d ops, %.2f%% size\n",$words,$nr_stems,$stems,($nr_stems*100/$words); |
82 |
|
|
83 |
|
foreach my $s (keys %stem_words) { |
84 |
|
print "#stem $stem_words{$s} $s\n"; |
85 |
|
} |
86 |
|
|
87 |
|
foreach my $r (keys %rules) { |
88 |
|
print "#rule $rules{$r} $r\n"; |
89 |
} |
} |