line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::YI::Romanize; |
2
|
3
|
|
|
3
|
|
1825
|
use utf8; |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
22
|
|
3
|
|
|
|
|
|
|
|
4
|
3
|
|
|
3
|
|
85
|
use strict; |
|
3
|
|
|
|
|
3
|
|
|
3
|
|
|
|
|
61
|
|
5
|
3
|
|
|
3
|
|
25
|
use warnings; |
|
3
|
|
|
|
|
10
|
|
|
3
|
|
|
|
|
147
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
our $VERSION = '0.01'; |
8
|
|
|
|
|
|
|
|
9
|
3
|
|
|
3
|
|
2060
|
use Unicode::Normalize; |
|
3
|
|
|
|
|
512556
|
|
|
3
|
|
|
|
|
3386
|
|
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
our $normalize_combinings; |
12
|
|
|
|
|
|
|
our $yivo2latn; |
13
|
|
|
|
|
|
|
our $vowels; |
14
|
|
|
|
|
|
|
our $consonants; |
15
|
|
|
|
|
|
|
our $consonants_2; |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
sub new { |
19
|
6
|
|
|
6
|
1
|
699
|
my $class = shift; |
20
|
|
|
|
|
|
|
# uncoverable condition false |
21
|
6
|
100
|
66
|
|
|
37
|
bless @_ ? @_ > 1 ? {@_} : {%{$_[0]}} : {}, ref $class || $class; |
|
2
|
100
|
|
|
|
11
|
|
22
|
|
|
|
|
|
|
} |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
sub _gen_charclass { |
25
|
12
|
|
|
12
|
|
13
|
my $chartable = shift; |
26
|
12
|
|
|
|
|
9
|
my $string; |
27
|
12
|
|
|
|
|
11
|
for my $cons (@{$chartable}) { |
|
12
|
|
|
|
|
68
|
|
28
|
180
|
|
|
|
|
192
|
$string .= $cons->[0]; |
29
|
|
|
|
|
|
|
} |
30
|
12
|
|
|
|
|
61
|
return join('|',split('',$string)); |
31
|
|
|
|
|
|
|
} |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
sub _normalize { |
34
|
4
|
|
|
4
|
|
123914
|
my $string = shift; |
35
|
4
|
|
|
|
|
17
|
for my $rule (@$normalize_combinings) { |
36
|
68
|
|
|
|
|
2074
|
$string =~ s/$rule->[0]/$rule->[1]/g; |
37
|
|
|
|
|
|
|
} |
38
|
4
|
|
|
|
|
12
|
return $string; |
39
|
|
|
|
|
|
|
} |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub yivo2latn { |
42
|
2
|
|
|
2
|
1
|
417
|
my ($self, $text) = @_; |
43
|
2
|
|
|
|
|
8
|
my $string = _normalize(NFC($text)); |
44
|
|
|
|
|
|
|
|
45
|
2
|
|
|
|
|
443
|
my $cons_2 = _gen_charclass($consonants_2);; |
46
|
2
|
|
|
|
|
6
|
my $cons = _gen_charclass($consonants); |
47
|
2
|
|
|
|
|
10
|
my $vowels = _gen_charclass($vowels); |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
# "\x{05D9}" HEBREW LETTER YOD|י|y, i |
50
|
|
|
|
|
|
|
# 1.1. y before or after a vowel; |
51
|
|
|
|
|
|
|
# i between consonants; |
52
|
|
|
|
|
|
|
# y after טt, דd, סs, זz, לl, נn; and before a vowel |
53
|
|
|
|
|
|
|
# indicates the palatals in words of Slavic origin. |
54
|
|
|
|
|
|
|
|
55
|
2
|
|
|
|
|
280
|
$string =~ s/\x{05D9}($vowels)/y$1/g; |
56
|
2
|
|
|
|
|
173
|
$string =~ s/($vowels)\x{05D9}/${1}y/g; |
57
|
2
|
|
|
|
|
151
|
$string =~ s/($cons_2)\x{05D9}/${1}y/g; |
58
|
2
|
|
|
|
|
329
|
$string =~ s/($cons)\x{05D9}($cons)/${1}i${2}/g; |
59
|
2
|
|
|
|
|
69
|
$string =~ s/\x{05D9}/i/g; |
60
|
|
|
|
|
|
|
|
61
|
2
|
|
|
|
|
25
|
for my $rule (@$yivo2latn) { |
62
|
88
|
|
|
|
|
2459
|
$string =~ s/$rule->[0]/$rule->[1]/g; |
63
|
|
|
|
|
|
|
} |
64
|
2
|
|
|
|
|
62
|
return $string; |
65
|
|
|
|
|
|
|
} |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
sub yivo2ipa { |
68
|
2
|
|
|
2
|
1
|
330
|
my ($self, $text) = @_; |
69
|
2
|
|
|
|
|
7
|
my $string = _normalize(NFC($text)); |
70
|
|
|
|
|
|
|
|
71
|
2
|
|
|
|
|
406
|
my $cons_2 = _gen_charclass($consonants_2);; |
72
|
2
|
|
|
|
|
5
|
my $cons = _gen_charclass($consonants); |
73
|
2
|
|
|
|
|
6
|
my $vowels = _gen_charclass($vowels); |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
# "\x{05D9}" HEBREW LETTER YOD|י|y, i |
76
|
|
|
|
|
|
|
# 1.1. y before or after a vowel; |
77
|
|
|
|
|
|
|
# i between consonants; |
78
|
|
|
|
|
|
|
# y after טt, דd, סs, זz, לl, נn; and before a vowel |
79
|
|
|
|
|
|
|
# indicates the palatals in words of Slavic origin. |
80
|
|
|
|
|
|
|
|
81
|
2
|
|
|
|
|
260
|
$string =~ s/\x{05D9}($vowels)/j$1/g; |
82
|
2
|
|
|
|
|
120
|
$string =~ s/($vowels)\x{05D9}/${1}j/g; |
83
|
2
|
|
|
|
|
276
|
$string =~ s/($cons)\x{05D9}($cons)/${1}i${2}/g; |
84
|
2
|
|
|
|
|
89
|
$string =~ s/($cons_2)\x{05D9}/${1}j/g; |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
|
87
|
2
|
|
|
|
|
8
|
for my $rule (@$yivo2latn) { |
88
|
88
|
|
|
|
|
2359
|
$string =~ s/$rule->[0]/$rule->[2]/g; |
89
|
|
|
|
|
|
|
} |
90
|
2
|
|
|
|
|
54
|
return $string; |
91
|
|
|
|
|
|
|
} |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
BEGIN { |
94
|
3
|
|
|
3
|
|
22
|
$normalize_combinings = [ |
95
|
|
|
|
|
|
|
["\x{05D0}\x{05B7}","\x{FB2E}"], # HEBREW LETTER ALEF WITH PATAH |
96
|
|
|
|
|
|
|
["\x{05D0}\x{05B8}","\x{FB2F}"], # HEBREW LETTER ALEF WITH QAMATS |
97
|
|
|
|
|
|
|
["\x{05D1}\x{05BC}","\x{FB31}"], # HEBREW LETTER BET WITH DAGESH |
98
|
|
|
|
|
|
|
["\x{05D1}\x{05BF}","\x{FB4C}"], # HEBREW LETTER BET WITH RAFE |
99
|
|
|
|
|
|
|
["\x{05D5}\x{05BC}","\x{FB35}"], # HEBREW LETTER VAV WITH DAGESH |
100
|
|
|
|
|
|
|
["\x{05D5}\x{05D5}","\x{05F0}"], # HEBREW LIGATURE YIDDISH DOUBLE VAV |
101
|
|
|
|
|
|
|
["\x{05D5}\x{05D9}","\x{05F1}"], # HEBREW LIGATURE YIDDISH VAV YOD |
102
|
|
|
|
|
|
|
["\x{05D9}\x{05B4}","\x{FB1D}"], # HEBREW LETTER YOD WITH HIRIQ |
103
|
|
|
|
|
|
|
["\x{05D9}\x{05D9}","\x{05F2}"], # HEBREW LIGATURE YIDDISH DOUBLE YOD |
104
|
|
|
|
|
|
|
["\x{05F2}\x{05B7}","\x{FB1F}"], # HEBREW LIGATURE YIDDISH YOD YOD PATAH |
105
|
|
|
|
|
|
|
["\x{05DB}\x{05BC}","\x{FB3B}"], # HEBREW LETTER KAF WITH DAGESH |
106
|
|
|
|
|
|
|
["\x{05E4}\x{05BC}","\x{FB44}"], # HEBREW LETTER PE WITH DAGESH |
107
|
|
|
|
|
|
|
["\x{05E4}\x{05BF}","\x{FB4E}"], # HEBREW LETTER PE WITH RAFE |
108
|
|
|
|
|
|
|
["\x{05E9}\x{05C2}","\x{FB2B}"], # HEBREW LETTER SHIN WITH SIN DOT |
109
|
|
|
|
|
|
|
["\x{05EA}\x{05BC}","\x{FB4A}"], # HEBREW LETTER TAV WITH DAGESH |
110
|
|
|
|
|
|
|
["\x{FB20}","\x{05E2}"], # HEBREW LETTER ALTERNATIVE AYIN |
111
|
|
|
|
|
|
|
["\x{05E4}","\x{05BF}"], # HEBREW LETTER PE |
112
|
|
|
|
|
|
|
]; |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
|
115
|
3
|
|
|
|
|
50
|
$yivo2latn = [ |
116
|
|
|
|
|
|
|
["\x{05D3}\x{05D6}\x{05E9}",'dzh','d͡ʒ'], |
117
|
|
|
|
|
|
|
["\x{05D6}\x{05E9}",'zh','ʒ'], |
118
|
|
|
|
|
|
|
["\x{05D8}\x{05E9}",'tsh','t͡ʃ'], |
119
|
|
|
|
|
|
|
["\x{05D0}",'',''], |
120
|
|
|
|
|
|
|
["\x{FB2E}",'a','a'], |
121
|
|
|
|
|
|
|
["\x{FB2F}",'o','ɔ'], |
122
|
|
|
|
|
|
|
["\x{05D1}",'b','b'], |
123
|
|
|
|
|
|
|
["\x{FB31}",'','b'], |
124
|
|
|
|
|
|
|
["\x{FB4C}",'v','v'], |
125
|
|
|
|
|
|
|
["\x{05D2}",'g','ɡ'], |
126
|
|
|
|
|
|
|
["\x{05D3}",'d','d'], |
127
|
|
|
|
|
|
|
["\x{05D4}",'h','h'], |
128
|
|
|
|
|
|
|
["\x{05D5}",'u','ʊ'], |
129
|
|
|
|
|
|
|
["\x{FB35}",'u','ʊ'], |
130
|
|
|
|
|
|
|
# "\x{05D5}" HEBREW LETTER VAV,"\x{05B9}" HEBREW POINT HOLAM|וֹ|(none)|(none)|ɔ, ɔj|(o,oj)|khoylem|Non-YIVO alternative to אָ and וי. |
131
|
|
|
|
|
|
|
["\x{05F0}",'v','v'], |
132
|
|
|
|
|
|
|
["\x{05F1}",'oy','ɔj'], |
133
|
|
|
|
|
|
|
["\x{05D6}",'z','z'], |
134
|
|
|
|
|
|
|
["\x{05D7}",'kh','x'], |
135
|
|
|
|
|
|
|
["\x{05D8}",'t','t'], |
136
|
|
|
|
|
|
|
# TODO "\x{05D9}" HEBREW LETTER YOD|י|y, i|y, i|j, i|j, i|yud|Consonantal [j] when the first character in a syllable. Vocalic [i] otherwise. |
137
|
|
|
|
|
|
|
["\x{FB1D}",'i','i'], |
138
|
|
|
|
|
|
|
["\x{05F2}",'ey','ɛj'], |
139
|
|
|
|
|
|
|
["\x{FB1F}",'ay','aj'], |
140
|
|
|
|
|
|
|
["\x{FB3B}",'k','k'], |
141
|
|
|
|
|
|
|
["\x{05DB}",'kh','x'], |
142
|
|
|
|
|
|
|
["\x{05DA}",'kh','x'], |
143
|
|
|
|
|
|
|
["\x{05DC}",'l','l'], # TODO: ʎ |
144
|
|
|
|
|
|
|
["\x{05DE}",'m','m'], |
145
|
|
|
|
|
|
|
["\x{05DD}",'m','m'], |
146
|
|
|
|
|
|
|
["\x{05E0}",'n','n'], |
147
|
|
|
|
|
|
|
["\x{05DF}",'n','n'], # TODO: ŋ, m |
148
|
|
|
|
|
|
|
["\x{05E1}",'s','s'], |
149
|
|
|
|
|
|
|
["\x{05E2}",'e','ɛ'], # TODO: ə |
150
|
|
|
|
|
|
|
["\x{FB44}",'p','p'], |
151
|
|
|
|
|
|
|
["\x{FB4E}",'f','f'], |
152
|
|
|
|
|
|
|
["\x{05E3}",'f','f'], |
153
|
|
|
|
|
|
|
["\x{05BF}",'f','f'], |
154
|
|
|
|
|
|
|
["\x{05E6}",'ts','ts'], |
155
|
|
|
|
|
|
|
["\x{05E5}",'ts','ts'], |
156
|
|
|
|
|
|
|
["\x{05E7}",'k','k'], |
157
|
|
|
|
|
|
|
["\x{05E8}",'r','ʀ'], |
158
|
|
|
|
|
|
|
["\x{05E9}",'sh','ʃ'], |
159
|
|
|
|
|
|
|
["\x{FB2B}",'s','s'], |
160
|
|
|
|
|
|
|
["\x{FB4A}",'t','t'], |
161
|
|
|
|
|
|
|
["\x{05EA}",'s','s'], |
162
|
|
|
|
|
|
|
]; |
163
|
|
|
|
|
|
|
|
164
|
3
|
|
|
|
|
22
|
$vowels = [ |
165
|
|
|
|
|
|
|
["\x{FB2E}",'a'], |
166
|
|
|
|
|
|
|
["\x{FB2F}",'o'], |
167
|
|
|
|
|
|
|
["\x{05D5}",'u'], |
168
|
|
|
|
|
|
|
["\x{FB35}",'u'], |
169
|
|
|
|
|
|
|
["\x{05F1}",'oy'], |
170
|
|
|
|
|
|
|
# TODO "\x{05D9}" HEBREW LETTER YOD|י|y, i|y, i|j, i|j, i|yud|Consonantal [j] when the first character in a syllable. Vocalic [i] otherwise. |
171
|
|
|
|
|
|
|
["\x{FB1D}",'i'], |
172
|
|
|
|
|
|
|
["\x{05F2}",'ey'], |
173
|
|
|
|
|
|
|
["\x{FB1F}",'ay'], |
174
|
|
|
|
|
|
|
["\x{05E2}",'e'], |
175
|
|
|
|
|
|
|
]; |
176
|
|
|
|
|
|
|
|
177
|
3
|
|
|
|
|
24
|
$consonants = [ |
178
|
|
|
|
|
|
|
["\x{05D1}",'b'], |
179
|
|
|
|
|
|
|
["\x{FB4C}",'v'], |
180
|
|
|
|
|
|
|
["\x{05D2}",'g'], |
181
|
|
|
|
|
|
|
["\x{05D3}",'d'], |
182
|
|
|
|
|
|
|
["\x{05D4}",'h'], |
183
|
|
|
|
|
|
|
["\x{05F0}",'v'], |
184
|
|
|
|
|
|
|
["\x{05D6}",'z'], |
185
|
|
|
|
|
|
|
["\x{05D7}",'kh'], |
186
|
|
|
|
|
|
|
["\x{05D8}",'t'], |
187
|
|
|
|
|
|
|
["\x{FB3B}",'k'], |
188
|
|
|
|
|
|
|
["\x{05DB}",'kh'], |
189
|
|
|
|
|
|
|
["\x{05DA}",'kh'], |
190
|
|
|
|
|
|
|
["\x{05DC}",'l'], |
191
|
|
|
|
|
|
|
["\x{05DE}",'m'], |
192
|
|
|
|
|
|
|
["\x{05DD}",'m'], |
193
|
|
|
|
|
|
|
["\x{05E0}",'n'], |
194
|
|
|
|
|
|
|
["\x{05DF}",'n'], |
195
|
|
|
|
|
|
|
["\x{05E1}",'s'], |
196
|
|
|
|
|
|
|
["\x{FB44}",'p'], |
197
|
|
|
|
|
|
|
["\x{FB4E}",'f'], |
198
|
|
|
|
|
|
|
["\x{05E3}",'f'], |
199
|
|
|
|
|
|
|
["\x{05BF}",'f'], |
200
|
|
|
|
|
|
|
["\x{05E6}",'ts'], |
201
|
|
|
|
|
|
|
["\x{05E5}",'ts'], |
202
|
|
|
|
|
|
|
["\x{05E7}",'k'], |
203
|
|
|
|
|
|
|
["\x{05E8}",'r'], |
204
|
|
|
|
|
|
|
["\x{05E9}",'sh'], |
205
|
|
|
|
|
|
|
["\x{FB2B}",'s'], |
206
|
|
|
|
|
|
|
["\x{FB4A}",'t'], |
207
|
|
|
|
|
|
|
["\x{05EA}",'s'], |
208
|
|
|
|
|
|
|
]; |
209
|
|
|
|
|
|
|
|
210
|
3
|
|
|
|
|
121
|
$consonants_2 = [ |
211
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
["\x{05D3}",'d'], |
213
|
|
|
|
|
|
|
["\x{05D6}",'z'], |
214
|
|
|
|
|
|
|
["\x{05D8}",'t'], |
215
|
|
|
|
|
|
|
["\x{05DC}",'l'], |
216
|
|
|
|
|
|
|
["\x{05E0}",'n'], |
217
|
|
|
|
|
|
|
["\x{05E1}",'s'], |
218
|
|
|
|
|
|
|
#["\x{FB4A}",'t'], |
219
|
|
|
|
|
|
|
]; |
220
|
|
|
|
|
|
|
} |
221
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
1; |
223
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
__END__ |