1 |
#!/usr/bin/perl -w |
2 |
# |
3 |
# Croatian stemmer |
4 |
# |
5 |
|
6 |
# promjenjive: |
7 |
# - imenice |
8 |
# - pridjevi |
9 |
# - brojevi |
10 |
# - zamjenice |
11 |
# - prilozi |
12 |
# - glagoli |
13 |
# |
14 |
# nepromjenjive: |
15 |
# - prijedlozi |
16 |
# - veznici |
17 |
# - èestice |
18 |
# - uzvici |
19 |
# |
20 |
|
21 |
sub rm_a { |
22 |
s/a//g; |
23 |
} |
24 |
|
25 |
sub kgh { |
26 |
my ($pre,$replace,$post) = @_; |
27 |
$replace =~ s/[cè]/k/g; |
28 |
$replace =~ s/[z¾]/g/g; |
29 |
$replace =~ s/[s¹]/h/g; |
30 |
return $pre . $replace . $post . "\tkgh"; |
31 |
} |
32 |
|
33 |
# samoglasnici |
34 |
my $sa = '[aeiou]'; |
35 |
# suglasnici |
36 |
my $su = '[^aeiou]'; |
37 |
|
38 |
while(<>) { |
39 |
chomp; |
40 |
|
41 |
my $orig = $_; |
42 |
|
43 |
# imenice |
44 |
# vrsta a |
45 |
|
46 |
unless ( |
47 |
# nepostojano a |
48 |
s/(${su}a{$su})a/$1 1/g || |
49 |
s/(${su})(${su})(a|u|i|e|om|ima)$/$1a$2\t3/g |
50 |
) { |
51 |
# kgh |
52 |
s/(${su}${sa})([è¾¹czs])(i|e|ima)$/kgh($1,$2,'')/gex; |
53 |
# s/${su}${sa}[c](i|ima) 4//g; |
54 |
s/(${su}${sa}[kgh])(a|u|om)\t5/$1/g; |
55 |
|
56 |
# mno¾ina |
57 |
s/(${su})(ov|ev)*(i|a|ima|e|in|ina)$/$1\t6/g; |
58 |
# jednina |
59 |
s/(o|e|a|u|om|em)$/\t7/g; |
60 |
} |
61 |
|
62 |
|
63 |
printf("%-15s %s\n",$orig,$_); |
64 |
} |