Commit 512dab52889ae2c8de833d7ba62865f34ae62711
1 parent
9e0e916e
łączenie zasad ortograficznych poszczególnych języków
Showing
7 changed files
with
773 additions
and
598 deletions
morphology/data/alternations.dic
morphology/data/fonetics_de.dic
... | ... | @@ -17,72 +17,72 @@ |
17 | 17 | ρ aε axε am′iε ąε eε ęε iε oε omε onε umε óvε uε ov′iε omε emε ε |
18 | 18 | |
19 | 19 | @rev_rules |
20 | -r rh | |
21 | -r r | |
22 | -ta tha | |
23 | -ks x | |
20 | +de r rh | |
21 | +de r r | |
22 | +de ta tha | |
23 | +de ks x | |
24 | 24 | |
25 | -g′e ge ε | |
26 | -k′e ke ε | |
25 | +de g′e ge ε | |
26 | +de k′e ke ε | |
27 | 27 | |
28 | -k′i ky ε | |
28 | +de k′i ky ε | |
29 | 29 | |
30 | 30 | |
31 | 31 | @rules |
32 | -š sch | |
33 | -r rh | |
34 | -c z | |
35 | -ał au | |
32 | +de š sch | |
33 | +de r rh | |
34 | +de c z | |
35 | +de ał au | |
36 | 36 | |
37 | -ře rze ε | |
38 | -ks′e ksie ε | |
39 | -t′e cie ε | |
40 | -v′e vie ε | |
37 | +de ře rze ε | |
38 | +de ks′e ksie ε | |
39 | +de t′e cie ε | |
40 | +de v′e vie ε | |
41 | 41 | |
42 | -óv ów ε | |
43 | -ov′i owi ε | |
44 | -ov′e owie ε | |
45 | -am′i ami ε | |
46 | -ax ach ε | |
42 | +de óv ów ε | |
43 | +de ov′i owi ε | |
44 | +de ov′e owie ε | |
45 | +de am′i ami ε | |
46 | +de ax ach ε | |
47 | 47 | |
48 | -g′e ge ch | |
49 | -g′e ge go | |
50 | -g′e ge m | |
51 | -g′e ge mi | |
52 | -g′e ge mu | |
53 | -k′e ke ch | |
54 | -k′e ke go | |
55 | -k′e ke m | |
56 | -k′e ke mi | |
57 | -k′e ke mu | |
58 | -k′i ky ch | |
59 | -k′i ky mi | |
60 | -k′i ky m | |
61 | -x ch ε | |
62 | -m′i mi ε | |
48 | +de g′e ge ch | |
49 | +de g′e ge go | |
50 | +de g′e ge m | |
51 | +de g′e ge mi | |
52 | +de g′e ge mu | |
53 | +de k′e ke ch | |
54 | +de k′e ke go | |
55 | +de k′e ke m | |
56 | +de k′e ke mi | |
57 | +de k′e ke mu | |
58 | +de k′i ky ch | |
59 | +de k′i ky mi | |
60 | +de k′i ky m | |
61 | +de x ch ε | |
62 | +de m′i mi ε | |
63 | 63 | |
64 | -kj ky ’ | |
65 | - ’ ρ | |
64 | +de kj ky ’ | |
65 | +de ’ ρ | |
66 | 66 | |
67 | -góv gów ε | |
68 | -gov′i gowi ε | |
69 | -gov′e gowie ε | |
70 | -gam′i gami ε | |
71 | -gax gach ε | |
72 | -gom gom ε | |
73 | -kóv ków ε | |
74 | -kov′i kowi ε | |
75 | -kov′e kowie ε | |
76 | -kam′i kami ε | |
77 | -kax kach ε | |
78 | -kom kom ε | |
67 | +de góv gów ε | |
68 | +de gov′i gowi ε | |
69 | +de gov′e gowie ε | |
70 | +de gam′i gami ε | |
71 | +de gax gach ε | |
72 | +de gom gom ε | |
73 | +de kóv ków ε | |
74 | +de kov′i kowi ε | |
75 | +de kov′e kowie ε | |
76 | +de kam′i kami ε | |
77 | +de kax kach ε | |
78 | +de kom kom ε | |
79 | 79 | |
80 | -k′ix kich ε | |
81 | -k′e kie ε | |
82 | -k′im kim ε | |
83 | -k′im′i kimi ε | |
84 | -k′ego kiego ε | |
85 | -k′emu kiemu ε | |
80 | +de k′ix kich ε | |
81 | +de k′e kie ε | |
82 | +de k′im kim ε | |
83 | +de k′im′i kimi ε | |
84 | +de k′ego kiego ε | |
85 | +de k′emu kiemu ε | |
86 | 86 | |
87 | 87 | |
88 | 88 | |
... | ... |
morphology/data/fonetics_en.dic
... | ... | @@ -17,196 +17,196 @@ |
17 | 17 | ρ aε axε am′iε ąε eε ęε iε oε omε onε umε óvε uε ov′iε omε emε ε |
18 | 18 | |
19 | 19 | @rev_rules |
20 | -ks x | |
21 | -č tch | |
22 | -ł w | |
23 | -aj ay ρ | |
24 | -ej ey ρ | |
25 | -oj oy ρ | |
26 | - | |
27 | -d de ε | |
28 | -d d ε | |
29 | -f ph ε | |
30 | -k ke ε | |
31 | -k k ε | |
32 | -k kes ε | |
33 | -k c | |
34 | -k ky ε | |
35 | -t te ε | |
36 | -t t ε | |
37 | -t th | |
38 | -t the ε | |
39 | -r re ε | |
40 | -r r ε | |
41 | -r rh ε | |
42 | -m me ε | |
43 | -m m ε | |
44 | -n ne ε | |
45 | -n n ε | |
46 | -s se ε | |
47 | -s s ε | |
48 | -s ce ε | |
49 | -s th ε | |
50 | -s the ε | |
51 | - | |
52 | -a a ε | |
53 | -a ah ε | |
54 | -ka kha ε | |
55 | -vja via ε | |
56 | - | |
57 | -ʲi ie ε | |
58 | -ʲi i ε | |
59 | -ʲi y ε | |
20 | +en ks x | |
21 | +en č tch | |
22 | +en ł w | |
23 | +en aj ay ρ | |
24 | +en ej ey ρ | |
25 | +en oj oy ρ | |
26 | + | |
27 | +en d de ε | |
28 | +en d d ε | |
29 | +en f ph ε | |
30 | +en k ke ε | |
31 | +en k k ε | |
32 | +en k kes ε | |
33 | +en k c | |
34 | +en k ky ε | |
35 | +en t te ε | |
36 | +en t t ε | |
37 | +en t th | |
38 | +en t the ε | |
39 | +en r re ε | |
40 | +en r r ε | |
41 | +en r rh ε | |
42 | +en m me ε | |
43 | +en m m ε | |
44 | +en n ne ε | |
45 | +en n n ε | |
46 | +en s se ε | |
47 | +en s s ε | |
48 | +en s ce ε | |
49 | +en s th ε | |
50 | +en s the ε | |
51 | + | |
52 | +en a a ε | |
53 | +en a ah ε | |
54 | +en ka kha ε | |
55 | +en vja via ε | |
56 | + | |
57 | +en ʲi ie ε | |
58 | +en ʲi i ε | |
59 | +en ʲi y ε | |
60 | 60 | |
61 | 61 | @rules |
62 | -ks x | |
63 | -č tch | |
64 | -ł w | |
65 | -d dh | |
66 | -g gh | |
67 | -k kh | |
68 | -l leigh | |
69 | -aj ay ρ | |
70 | -ej ey ρ | |
71 | -oj oy ρ | |
72 | -aʲi ai ε | |
73 | -oʲi oi ε | |
74 | -eʲi ei ε | |
75 | -uʲi ui ε | |
76 | - | |
77 | -vj vi ρ | |
78 | -v′ij vij ε | |
79 | - | |
80 | -ce ce ε | |
81 | -f′e fie ε | |
82 | -ře rze ε | |
83 | -t′e cie ε | |
84 | -kt′e kcie ε | |
85 | -s′t′e ście ε | |
86 | -m′e mie ε | |
87 | -n′e nie ε | |
88 | -d′e dzie ε | |
89 | -p′e pie ε | |
90 | -s′e sie ε | |
91 | -ks′e ksie ε | |
92 | -v′e vie ε | |
93 | -z′e zie ε | |
94 | - | |
95 | -k′i ki ε | |
96 | -k′em kiem ε | |
97 | -k k iem | |
98 | -řy rzy ε | |
99 | - | |
100 | - | |
101 | -ʲi ie ch | |
102 | -ʲi ie mi | |
103 | -ʲi ie m | |
104 | -ʲi ee ch | |
105 | -ʲi ee mi | |
106 | -li ly ch | |
107 | -li ly mi | |
108 | -li ly m | |
109 | -ʲego iego ε | |
110 | -ʲemu iemu ε | |
111 | -ʲem iem ε | |
112 | -ʲim im ε | |
113 | -ʲix ich ε | |
114 | -ʲe ie ε | |
115 | -ʲim′i imi ε | |
116 | -ʲov′e iowie ε | |
117 | -′e ie m | |
118 | -′i i ε | |
119 | -i i ε | |
120 | -x ch ε | |
121 | -m′i mi ε | |
122 | -óv ów ε | |
123 | -ov′i owi ε | |
124 | -ov′e owie ε | |
125 | -am′i ami ε | |
126 | -ax ach ε | |
127 | - | |
128 | -oł oe ’ | |
129 | -b by ’ | |
130 | -k′ ky ’ | |
131 | -r ry ’ | |
132 | -t thy ’ | |
133 | -d dy ’ | |
134 | -m my ’ | |
135 | -s cy ’ | |
136 | -l ly ’ | |
137 | -d dy ’ | |
138 | -s ce ’ | |
139 | -d de ’ | |
140 | -f fe ’ | |
141 | -ǯ ge ’ | |
142 | -k ke ’ | |
143 | -l le ’ | |
144 | -m me ’ | |
145 | -n ne ’ | |
146 | -r re ’ | |
147 | -s se ’ | |
148 | -t te ’ | |
149 | -t the ’ | |
150 | -v ve ’ | |
151 | -ł we ’ | |
152 | - ’ ρ | |
153 | - | |
154 | -s ce ε | |
155 | -d de ε | |
156 | -f fe ε | |
157 | -ǯ ge ε | |
158 | -k ke ε | |
159 | -l le ε | |
160 | -le le ε | |
161 | -m me ε | |
162 | -me me ε | |
163 | -n ne ε | |
164 | -r re ε | |
165 | -re re ε | |
166 | -s se ε | |
167 | -t te ε | |
168 | -v ve ε | |
169 | -ł we ε | |
170 | - | |
171 | - | |
172 | -#os′t′ ość ε | |
173 | -#os′t′ą ością ε | |
174 | -#os′t′om ościom ε | |
175 | -#os′t′am′i ościami ε | |
176 | -#os′t′ax ościach ε | |
177 | -#os′t′i ości ε | |
178 | -#ovą ową ε | |
179 | -#ovym′i owymi ε | |
180 | -#ovym owym ε | |
181 | -#ovyx owych ε | |
182 | -#ovy owy ε | |
183 | -#ovo owo ε | |
184 | -#ov′i owi ε | |
185 | -#ovemu owemu ε | |
186 | -#ovej owej ε | |
187 | -#ovego owego ε | |
188 | -#ove owe ε | |
189 | -#ova owa ε | |
190 | -#oły owy ε | |
191 | -#oła owa ε | |
192 | -#k′ix kich ε | |
193 | -#k′e kie ε | |
194 | -#k′ej kiej ε | |
195 | -#k′im kim ε | |
196 | -#k′im′i kimi ε | |
197 | -#ka ka ε | |
198 | -#ką ką ε | |
199 | -#ko ko ε | |
200 | -#ku ku ε | |
201 | -#če cze ε | |
202 | -#n′ja nia ε | |
203 | -#n′ją nią ε | |
204 | -#n′je nie ε | |
205 | -#n′ję nię ε | |
206 | -#n′ji nii ε | |
207 | -#n′ij nij ε | |
208 | -#n′jom niom ε | |
209 | -#n′jo nio ε | |
210 | -#n′jax niach ε | |
211 | -#n′jam′i niami ε | |
62 | +en ks x | |
63 | +en č tch | |
64 | +en ł w | |
65 | +en d dh | |
66 | +en g gh | |
67 | +en k kh | |
68 | +en l leigh | |
69 | +en aj ay ρ | |
70 | +en ej ey ρ | |
71 | +en oj oy ρ | |
72 | +en aʲi ai ε | |
73 | +en oʲi oi ε | |
74 | +en eʲi ei ε | |
75 | +en uʲi ui ε | |
76 | + | |
77 | +en vj vi ρ | |
78 | +en v′ij vij ε | |
79 | + | |
80 | +en ce ce ε | |
81 | +en f′e fie ε | |
82 | +en ře rze ε | |
83 | +en t′e cie ε | |
84 | +en kt′e kcie ε | |
85 | +en s′t′e ście ε | |
86 | +en m′e mie ε | |
87 | +en n′e nie ε | |
88 | +en d′e dzie ε | |
89 | +en p′e pie ε | |
90 | +en s′e sie ε | |
91 | +en ks′e ksie ε | |
92 | +en v′e vie ε | |
93 | +en z′e zie ε | |
94 | + | |
95 | +en k′i ki ε | |
96 | +en k′em kiem ε | |
97 | +en k k iem | |
98 | +en řy rzy ε | |
99 | + | |
100 | + | |
101 | +en ʲi ie ch | |
102 | +en ʲi ie mi | |
103 | +en ʲi ie m | |
104 | +en ʲi ee ch | |
105 | +en ʲi ee mi | |
106 | +en li ly ch | |
107 | +en li ly mi | |
108 | +en li ly m | |
109 | +en ʲego iego ε | |
110 | +en ʲemu iemu ε | |
111 | +en ʲem iem ε | |
112 | +en ʲim im ε | |
113 | +en ʲix ich ε | |
114 | +en ʲe ie ε | |
115 | +en ʲim′i imi ε | |
116 | +en ʲov′e iowie ε | |
117 | +en ′e ie m | |
118 | +en ′i i ε | |
119 | +en i i ε | |
120 | +en x ch ε | |
121 | +en m′i mi ε | |
122 | +en óv ów ε | |
123 | +en ov′i owi ε | |
124 | +en ov′e owie ε | |
125 | +en am′i ami ε | |
126 | +en ax ach ε | |
127 | + | |
128 | +en oł oe ’ | |
129 | +en b by ’ | |
130 | +en k′ ky ’ | |
131 | +en r ry ’ | |
132 | +en t thy ’ | |
133 | +en d dy ’ | |
134 | +en m my ’ | |
135 | +en s cy ’ | |
136 | +en l ly ’ | |
137 | +en d dy ’ | |
138 | +en s ce ’ | |
139 | +en d de ’ | |
140 | +en f fe ’ | |
141 | +en ǯ ge ’ | |
142 | +en k ke ’ | |
143 | +en l le ’ | |
144 | +en m me ’ | |
145 | +en n ne ’ | |
146 | +en r re ’ | |
147 | +en s se ’ | |
148 | +en t te ’ | |
149 | +en t the ’ | |
150 | +en v ve ’ | |
151 | +en ł we ’ | |
152 | +en ’ ρ | |
153 | + | |
154 | +en s ce ε | |
155 | +en d de ε | |
156 | +en f fe ε | |
157 | +en ǯ ge ε | |
158 | +en k ke ε | |
159 | +en l le ε | |
160 | +en le le ε | |
161 | +en m me ε | |
162 | +en me me ε | |
163 | +en n ne ε | |
164 | +en r re ε | |
165 | +en re re ε | |
166 | +en s se ε | |
167 | +en t te ε | |
168 | +en v ve ε | |
169 | +en ł we ε | |
170 | + | |
171 | + | |
172 | +#en os′t′ ość ε | |
173 | +#en os′t′ą ością ε | |
174 | +#en os′t′om ościom ε | |
175 | +#en os′t′am′i ościami ε | |
176 | +#en os′t′ax ościach ε | |
177 | +#en os′t′i ości ε | |
178 | +#en ovą ową ε | |
179 | +#en ovym′i owymi ε | |
180 | +#en ovym owym ε | |
181 | +#en ovyx owych ε | |
182 | +#en ovy owy ε | |
183 | +#en ovo owo ε | |
184 | +#en ov′i owi ε | |
185 | +#en ovemu owemu ε | |
186 | +#en ovej owej ε | |
187 | +#en ovego owego ε | |
188 | +#en ove owe ε | |
189 | +#en ova owa ε | |
190 | +#en oły owy ε | |
191 | +#en oła owa ε | |
192 | +#en k′ix kich ε | |
193 | +#en k′e kie ε | |
194 | +#en k′ej kiej ε | |
195 | +#en k′im kim ε | |
196 | +#en k′im′i kimi ε | |
197 | +#en ka ka ε | |
198 | +#en ką ką ε | |
199 | +#en ko ko ε | |
200 | +#en ku ku ε | |
201 | +#en če cze ε | |
202 | +#en n′ja nia ε | |
203 | +#en n′ją nią ε | |
204 | +#en n′je nie ε | |
205 | +#en n′ję nię ε | |
206 | +#en n′ji nii ε | |
207 | +#en n′ij nij ε | |
208 | +#en n′jom niom ε | |
209 | +#en n′jo nio ε | |
210 | +#en n′jax niach ε | |
211 | +#en n′jam′i niami ε | |
212 | 212 | |
... | ... |
morphology/data/fonetics_fr.dic
... | ... | @@ -17,105 +17,105 @@ |
17 | 17 | ρ aε axε am′iε ąε eε ęε iε oε omε onε umε óvε uε ov′iε omε emε ε |
18 | 18 | |
19 | 19 | @rev_rules |
20 | -ks x | |
21 | -b bes ε | |
22 | -d d ε | |
23 | -d de ε | |
24 | -f phe ε | |
25 | -g gue ε | |
26 | -g gues ε | |
27 | -k c ε | |
28 | -k que ε | |
29 | -k q ε | |
30 | -k cq ε | |
31 | -k ques ε | |
32 | -k cques ε | |
33 | -m me ε | |
34 | -n n ε | |
35 | -n ne ε | |
36 | -n nes ε | |
37 | -n′ ng ε | |
38 | -r r ε | |
39 | -r re ε | |
40 | -r res ε | |
41 | -r rs ε | |
42 | -s s ε | |
43 | -s se ε | |
44 | -s ce ε | |
45 | -t t ε | |
46 | -t tes ε | |
47 | -t thes ε | |
48 | -z se ε | |
20 | +fr ks x | |
21 | +fr b bes ε | |
22 | +fr d d ε | |
23 | +fr d de ε | |
24 | +fr f phe ε | |
25 | +fr g gue ε | |
26 | +fr g gues ε | |
27 | +fr k c ε | |
28 | +fr k que ε | |
29 | +fr k q ε | |
30 | +fr k cq ε | |
31 | +fr k ques ε | |
32 | +fr k cques ε | |
33 | +fr m me ε | |
34 | +fr n n ε | |
35 | +fr n ne ε | |
36 | +fr n nes ε | |
37 | +fr n′ ng ε | |
38 | +fr r r ε | |
39 | +fr r re ε | |
40 | +fr r res ε | |
41 | +fr r rs ε | |
42 | +fr s s ε | |
43 | +fr s se ε | |
44 | +fr s ce ε | |
45 | +fr t t ε | |
46 | +fr t tes ε | |
47 | +fr t thes ε | |
48 | +fr z se ε | |
49 | 49 | |
50 | 50 | @rules |
51 | -k q | |
52 | -n ng | |
51 | +fr k q | |
52 | +fr n ng | |
53 | 53 | |
54 | -b′e bie ε | |
55 | -d′e dzie ε | |
56 | -f′e fie ε | |
57 | -m′e mie ε | |
58 | -n′e nie ε | |
59 | -s′n′e śnie ε | |
60 | -ře rze ε | |
61 | -s′e sie ε | |
62 | -ks′e ksie ε | |
63 | -t′e cie ε | |
64 | -s′t′e ście ε | |
65 | -v′e vie ε | |
66 | -z′e zie ε | |
54 | +fr b′e bie ε | |
55 | +fr d′e dzie ε | |
56 | +fr f′e fie ε | |
57 | +fr m′e mie ε | |
58 | +fr n′e nie ε | |
59 | +fr s′n′e śnie ε | |
60 | +fr ře rze ε | |
61 | +fr s′e sie ε | |
62 | +fr ks′e ksie ε | |
63 | +fr t′e cie ε | |
64 | +fr s′t′e ście ε | |
65 | +fr v′e vie ε | |
66 | +fr z′e zie ε | |
67 | 67 | |
68 | -řy rzy ε | |
69 | -t′i ci ε | |
70 | -v′i vi ε | |
68 | +fr řy rzy ε | |
69 | +fr t′i ci ε | |
70 | +fr v′i vi ε | |
71 | 71 | |
72 | -g′em giem ε | |
73 | -g′i gi ε | |
74 | -k′em kiem ε | |
75 | -k′i ki ε | |
72 | +fr g′em giem ε | |
73 | +fr g′i gi ε | |
74 | +fr k′em kiem ε | |
75 | +fr k′i ki ε | |
76 | 76 | |
77 | -ʲi ieu ch | |
78 | -ʲe ieu go | |
79 | -ʲi ieu m | |
80 | -ʲi ieu mi | |
81 | -ʲe ieu mu | |
82 | -llj lli ego | |
83 | -llj lli emu | |
84 | -llj lli e | |
85 | -e ée go | |
86 | -e ée m | |
87 | -e ée mu | |
88 | -x ch ε | |
89 | -m′i mi ε | |
77 | +fr ʲi ieu ch | |
78 | +fr ʲe ieu go | |
79 | +fr ʲi ieu m | |
80 | +fr ʲi ieu mi | |
81 | +fr ʲe ieu mu | |
82 | +fr llj lli ego | |
83 | +fr llj lli emu | |
84 | +fr llj lli e | |
85 | +fr e ée go | |
86 | +fr e ée m | |
87 | +fr e ée mu | |
88 | +fr x ch ε | |
89 | +fr m′i mi ε | |
90 | 90 | |
91 | -óv ów ε | |
92 | -ov′i owi ε | |
93 | -ov′e owie ε | |
94 | -am′i ami ε | |
95 | -ax ach ε | |
91 | +fr óv ów ε | |
92 | +fr ov′i owi ε | |
93 | +fr ov′e owie ε | |
94 | +fr am′i ami ε | |
95 | +fr ax ach ε | |
96 | 96 | |
97 | -e ès ’ | |
98 | -e ais ’ | |
99 | -k que ’ | |
100 | -f phe ’ | |
101 | -sj sy ’ | |
102 | -fj phy ’ | |
103 | -rj ry ’ | |
104 | -š che ’ | |
105 | -d de ’ | |
106 | -r re ’ | |
107 | -z se ’ | |
108 | -ž ge ’ | |
109 | -# ine ’ | |
110 | -t te ’ | |
111 | -l le ’ | |
112 | -m me ’ | |
113 | -n′ gne ’ | |
114 | -n ne ’ | |
115 | -v ve ’ | |
116 | -g gue ’ | |
117 | -s ce ’ | |
118 | -gj guy ’ | |
119 | - ’ ρ | |
97 | +fr e ès ’ | |
98 | +fr e ais ’ | |
99 | +fr k que ’ | |
100 | +fr f phe ’ | |
101 | +fr sj sy ’ | |
102 | +fr fj phy ’ | |
103 | +fr rj ry ’ | |
104 | +fr š che ’ | |
105 | +fr d de ’ | |
106 | +fr r re ’ | |
107 | +fr z se ’ | |
108 | +fr ž ge ’ | |
109 | +fr # ine ’ | |
110 | +fr t te ’ | |
111 | +fr l le ’ | |
112 | +fr m me ’ | |
113 | +fr n′ gne ’ | |
114 | +fr n ne ’ | |
115 | +fr v ve ’ | |
116 | +fr g gue ’ | |
117 | +fr s ce ’ | |
118 | +fr gj guy ’ | |
119 | +fr ’ ρ | |
120 | 120 | |
121 | 121 | |
... | ... |
morphology/data/fonetics_pl.dic
... | ... | @@ -7,7 +7,7 @@ |
7 | 7 | λ b c ć d f g h k l ł m n ń p r s ś t w z ź ż - ε |
8 | 8 | κ b c ć d f g h i j k l ł m n ń p r s ś t w z ź ż - ε |
9 | 9 | σ a ą b c ć d e ę f g j h k l ł m n ń o ó p r s ś t u w y z ź ż - ε |
10 | -ρ aε achε amiε ąε eε ęε iε oε omε onε umε | |
10 | +ρ aε achε amiε ąε eε ęε iε oε omε onε umε ówε uε owiε omε emε yε owieε emuε egoε goε mε muε ε | |
11 | 11 | |
12 | 12 | @rev_symbols |
13 | 13 | ω a ą e ę o ó u |
... | ... | @@ -17,262 +17,427 @@ |
17 | 17 | ρ aε axε am′iε ąε eε ęε iε oε omε onε umε |
18 | 18 | |
19 | 19 | @rev_rules |
20 | -p′ pi ω | |
21 | -p′ p iκ | |
22 | -pj pi ρ | |
23 | -b′ bi ω | |
24 | -b′ b iκ | |
25 | -bj bi ρ | |
26 | -m′ mi ω | |
27 | -m′ m iκ | |
28 | -m′j mi ρ | |
29 | -f′ fi ω | |
30 | -f′ f iκ | |
31 | -fj fi ρ | |
32 | -v′ wi ω | |
33 | -v′ w iκ | |
34 | -v w δ | |
35 | -vj wi ρ | |
36 | -#v w j | |
37 | -t′ ć λ | |
38 | -t′ ci ω | |
39 | -t′ c iκ | |
40 | -d′ dź λ | |
41 | -d′ dzi ω | |
42 | -d′ dz iκ | |
43 | -ʒ dz | |
44 | -tʲ ti ω | |
45 | -tʲ t iκ | |
46 | -tj ti ρ | |
47 | -dʲ di ω | |
48 | -dʲ d iκ | |
49 | -dj di ρ | |
50 | -č cz δ | |
51 | -čʲ czi ω | |
52 | -čʲ cz iκ | |
53 | -č cz j | |
54 | -ǯ dż δ | |
55 | -ǯʲ dżi ω | |
56 | -ǯʲ dż iκ | |
57 | -ǯj dżi ρ | |
58 | -s′ ś λ | |
59 | -s′ si ω | |
60 | -s′ s iκ | |
61 | -z′ ź λ | |
62 | -z′ zi ω | |
63 | -z′ z iκ | |
64 | -š sz δ | |
65 | -š sz j | |
66 | -šʲ szi ω | |
67 | -šʲ sz iκ | |
68 | -ž ż δ | |
69 | -žʲ żi ω | |
70 | -žʲ ż iκ | |
71 | -žj żi ρ | |
72 | -lʲ li ω | |
73 | -lj li ρ | |
74 | -ř rz δ | |
75 | -ř rz j | |
76 | -rʲ ri ω | |
77 | -rʲ r iκ | |
78 | -rj ri ρ | |
79 | -n′ ń λ | |
80 | -n′ ni ω | |
81 | -n′ n iκ | |
82 | -n′j ni ρ | |
83 | -k′ ki ω | |
84 | -k′ k iκ | |
85 | -k′j ki ρ | |
86 | -g′ gi ω | |
87 | -g′ g iκ | |
88 | -g′j gi ρ | |
89 | -x ch δ | |
90 | -xʲ chi ω | |
91 | -xʲ ch iκ | |
92 | -xʲj chi ρ | |
93 | -hʲ hi ω | |
94 | -hʲ h iκ | |
95 | -aʲ a i | |
96 | -eʲ e i | |
97 | -oʲ o i | |
98 | -óʲ ó i | |
99 | -uʲ u i | |
100 | -iʲ i i | |
101 | -yʲ y i | |
20 | +pl p′ pi ω | |
21 | +pl p′ p iκ | |
22 | +pl pj pi ρ | |
23 | +pl b′ bi ω | |
24 | +pl b′ b iκ | |
25 | +pl bj bi ρ | |
26 | +pl m′ mi ω | |
27 | +pl m′ m iκ | |
28 | +pl m′j mi ρ | |
29 | +pl f′ fi ω | |
30 | +pl f′ f iκ | |
31 | +pl fj fi ρ | |
32 | +pl v′ wi ω | |
33 | +pl v′ w iκ | |
34 | +pl v w δ | |
35 | +pl vj wi ρ | |
36 | +#pl v w j | |
37 | +pl t′ ć λ | |
38 | +pl t′ ci ω | |
39 | +pl t′ c iκ | |
40 | +pl d′ dź λ | |
41 | +pl d′ dzi ω | |
42 | +pl d′ dz iκ | |
43 | +pl ʒ dz | |
44 | +pl tʲ ti ω | |
45 | +pl tʲ t iκ | |
46 | +pl tj ti ρ | |
47 | +pl dʲ di ω | |
48 | +pl dʲ d iκ | |
49 | +pl dj di ρ | |
50 | +pl č cz δ | |
51 | +pl čʲ czi ω | |
52 | +pl čʲ cz iκ | |
53 | +pl č cz j | |
54 | +pl ǯ dż δ | |
55 | +pl ǯʲ dżi ω | |
56 | +pl ǯʲ dż iκ | |
57 | +pl ǯj dżi ρ | |
58 | +pl s′ ś λ | |
59 | +pl s′ si ω | |
60 | +pl s′ s iκ | |
61 | +pl z′ ź λ | |
62 | +pl z′ zi ω | |
63 | +pl z′ z iκ | |
64 | +pl š sz δ | |
65 | +pl š sz j | |
66 | +pl šʲ szi ω | |
67 | +pl šʲ sz iκ | |
68 | +pl ž ż δ | |
69 | +pl žʲ żi ω | |
70 | +pl žʲ ż iκ | |
71 | +pl žj żi ρ | |
72 | +pl lʲ li ω | |
73 | +pl lj li ρ | |
74 | +pl ř rz δ | |
75 | +pl ř rz j | |
76 | +pl rʲ ri ω | |
77 | +pl rʲ r iκ | |
78 | +pl rj ri ρ | |
79 | +pl n′ ń λ | |
80 | +pl n′ ni ω | |
81 | +pl n′ n iκ | |
82 | +pl n′j ni ρ | |
83 | +pl k′ ki ω | |
84 | +pl k′ k iκ | |
85 | +pl k′j ki ρ | |
86 | +pl g′ gi ω | |
87 | +pl g′ g iκ | |
88 | +pl g′j gi ρ | |
89 | +pl x ch δ | |
90 | +pl xʲ chi ω | |
91 | +pl xʲ ch iκ | |
92 | +pl xʲj chi ρ | |
93 | +pl hʲ hi ω | |
94 | +pl hʲ h iκ | |
95 | +pl aʲ a i | |
96 | +pl eʲ e i | |
97 | +pl oʲ o i | |
98 | +pl óʲ ó i | |
99 | +pl uʲ u i | |
100 | +pl iʲ i i | |
101 | +pl yʲ y i | |
102 | + | |
103 | +de {sch}š sch | |
104 | +de {rh}r rh | |
105 | +de {z}c z | |
106 | +de {au}ał au | |
107 | +fr {q}k q | |
108 | +fr {nq}n ng | |
109 | +fr {qui}kʲ qui | |
110 | +de-fr {x}ks x | |
111 | +de r rh | |
112 | +de r r | |
113 | +de ta tha ε | |
114 | + t t | |
115 | +de-fr ks x | |
116 | +de {v}f v ε | |
117 | +fr {v}v v | |
118 | + | |
119 | +de g′e ge ε | |
120 | +de k′e ke ε | |
121 | +de g′el gel ε | |
122 | + | |
123 | +de k′i ky ε | |
124 | + | |
125 | +fr b bes ε | |
126 | +fr d d ε | |
127 | +fr d de ε | |
128 | +fr f phe ε | |
129 | +fr g gue ε | |
130 | +fr g gues ε | |
131 | +fr k c ε | |
132 | +fr k que ε | |
133 | +fr k q ε | |
134 | +fr k cq ε | |
135 | +fr k ques ε | |
136 | +fr k cques ε | |
137 | +fr m me ε | |
138 | +fr n n ε | |
139 | +fr n ne ε | |
140 | +fr n nes ε | |
141 | +fr n ng ε | |
142 | +fr r r ε | |
143 | +fr r re ε | |
144 | +fr r res ε | |
145 | +fr r rs ε | |
146 | +fr s s ε | |
147 | +fr s se ε | |
148 | +fr s ce ε | |
149 | +fr t t ε | |
150 | +fr t tes ε | |
151 | +fr t thes ε | |
152 | +fr z se ε | |
153 | + | |
154 | +de {ky}kj ky ε | |
155 | +fr {ès}e ès ε | |
156 | +fr {ais}e ais ε | |
157 | +fr {que}k que ε | |
158 | +fr {phe}f phe ε | |
159 | +fr {sy}sj sy ε | |
160 | +fr {phy}fj phy ε | |
161 | +fr {ry}rj ry ε | |
162 | +fr {che}š che ε | |
163 | +fr {de}d de ε | |
164 | +fr {re}r re ε | |
165 | +fr {se}z se ε | |
166 | +fr {ge}ž ge ε | |
167 | +fr {te}t te ε | |
168 | +fr {le}l le ε | |
169 | +fr {me}m me ε | |
170 | +fr {gne}n′ gne ε | |
171 | +fr {ne}n ne ε | |
172 | +#fr ine ε | |
173 | +fr {ve}v ve ε | |
174 | +fr {gue}g gue ε | |
175 | +fr {ce}s ce ε | |
176 | +fr {guy}gj guy ε | |
177 | + | |
178 | +fr {li}lj li ε | |
102 | 179 | |
103 | 180 | @rules |
104 | -p′ pi ω | |
105 | -p′ p iκ | |
106 | -#p p δ | |
107 | -#p p j | |
108 | -pj pi ρ | |
109 | -b′ bi ω | |
110 | -b′ b iκ | |
111 | -#b b δ | |
112 | -bj bi ρ | |
181 | +pl p′ pi ω | |
182 | +pl p′ p iκ | |
183 | +#pl p p δ | |
184 | +#pl p p j | |
185 | +pl pj pi ρ | |
186 | +pl b′ bi ω | |
187 | +pl b′ b iκ | |
188 | +#pl b b δ | |
189 | +pl bj bi ρ | |
113 | 190 | #zbitka |
114 | -#b b j | |
115 | -m′ mi ω | |
116 | -m′ m iκ | |
117 | -#m m δ | |
118 | -m′j mi ρ | |
119 | -f′ fi ω | |
120 | -f′ f iκ | |
121 | -#f f δ | |
122 | -fj fi ρ | |
123 | -v′ wi ω | |
124 | -v′ w iκ | |
125 | -v w δ | |
126 | -vj wi ρ | |
191 | +#pl b b j | |
192 | +pl m′ mi ω | |
193 | +pl m′ m iκ | |
194 | +#pl m m δ | |
195 | +pl m′j mi ρ | |
196 | +pl f′ fi ω | |
197 | +pl f′ f iκ | |
198 | +#pl f f δ | |
199 | +pl fj fi ρ | |
200 | +pl v′ wi ω | |
201 | +pl v′ w iκ | |
202 | +pl v w δ | |
203 | +pl vj wi ρ | |
127 | 204 | #zbitka |
128 | -#v w j | |
129 | -t′ ć λ | |
205 | +#pl v w j | |
206 | +pl t′ ć λ | |
130 | 207 | #zbitki |
131 | -#t′ ć u | |
132 | -#t′ ć a | |
133 | -#t′ ć i | |
134 | -t′ ci ω | |
135 | -#c c j | |
136 | -t′ c iκ | |
137 | -#c c μ | |
138 | -d′ dź λ | |
139 | -d′ dzi ω | |
140 | -d′ dz iκ | |
141 | -ʒ dz δ | |
208 | +#pl t′ ć u | |
209 | +#pl t′ ć a | |
210 | +#pl t′ ć i | |
211 | +pl t′ ci ω | |
212 | +#pl c c j | |
213 | +pl t′ c iκ | |
214 | +#pl c c μ | |
215 | +pl d′ dź λ | |
216 | +pl d′ dzi ω | |
217 | +pl d′ dz iκ | |
218 | +pl ʒ dz δ | |
142 | 219 | #zbitka |
143 | -#ʒ dz j | |
144 | -#t t δ | |
145 | -#t t j | |
146 | -tʲ ti ω | |
147 | -tʲ t iκ | |
148 | -tj ti ρ | |
149 | -#d d ξ | |
150 | -dʲ di ω | |
151 | -dʲ d iκ | |
152 | -#d d j | |
153 | -dj di ρ | |
154 | -č cz δ | |
155 | -čʲ czi ω | |
156 | -čʲ cz iκ | |
157 | -#čʲ cz j | |
158 | -ǯ dż δ | |
159 | -ǯʲ dżi ω | |
160 | -ǯʲ dż iκ | |
161 | -ǯj dżi ρ | |
162 | -s′ ś λ | |
163 | -s′ si ω | |
164 | -#s s j | |
165 | -s′ s iκ | |
166 | -#s s γ | |
167 | -z′ ź λ | |
168 | -z′ zi ω | |
169 | -#z z j | |
170 | -z′ z iκ | |
171 | -#z z δ | |
172 | -š sz δ | |
173 | -#šʲ sz j | |
174 | -šʲ szi ω | |
175 | -šʲ sz iκ | |
176 | -ž ż δ | |
177 | -žʲ żi ω | |
178 | -žʲ ż iκ | |
179 | -žj żi ρ | |
180 | -#l l δ | |
181 | -lʲ li ω | |
182 | -#l l iκ | |
183 | -#l l j | |
184 | -#ł ł δ | |
220 | +#pl ʒ dz j | |
221 | +#pl t t δ | |
222 | +#pl t t j | |
223 | +pl tʲ ti ω | |
224 | +pl tʲ t iκ | |
225 | +pl tj ti ρ | |
226 | +#pl d d ξ | |
227 | +pl dʲ di ω | |
228 | +pl dʲ d iκ | |
229 | +#pl d d j | |
230 | +pl dj di ρ | |
231 | +pl č cz δ | |
232 | +pl čʲ czi ω | |
233 | +pl čʲ cz iκ | |
234 | +#pl čʲ cz j | |
235 | +pl ǯ dż δ | |
236 | +pl ǯʲ dżi ω | |
237 | +pl ǯʲ dż iκ | |
238 | +pl ǯj dżi ρ | |
239 | +pl s′ ś λ | |
240 | +pl s′ si ω | |
241 | +#pl s s j | |
242 | +pl s′ s iκ | |
243 | +#pl s s γ | |
244 | +pl z′ ź λ | |
245 | +pl z′ zi ω | |
246 | +#pl z z j | |
247 | +pl z′ z iκ | |
248 | +#pl z z δ | |
249 | +pl š sz δ | |
250 | +#pl šʲ sz j | |
251 | +pl šʲ szi ω | |
252 | +pl šʲ sz iκ | |
253 | +pl ž ż δ | |
254 | +pl žʲ żi ω | |
255 | +pl žʲ ż iκ | |
256 | +pl žj żi ρ | |
257 | +#pl l l δ | |
258 | +pl lʲ li ω | |
259 | +pl l l iκ | |
260 | +#pl l l j | |
261 | +#pl ł ł δ | |
185 | 262 | #zbitka |
186 | -#ł ł i | |
263 | +#pl ł ł i | |
187 | 264 | #zbitka |
188 | -#ł ł j | |
189 | -lj li ρ | |
190 | -ř rz δ | |
265 | +#pl ł ł j | |
266 | +pl lj li ρ | |
267 | +pl ř rz δ | |
191 | 268 | #zbitka |
192 | -#ř rz j | |
193 | -r r zi | |
194 | -#r r γ | |
195 | -#r r j | |
196 | -rʲ ri ω | |
197 | -rʲ r iκ | |
198 | -rj ri ρ | |
199 | -n′ ń λ | |
200 | -n′ ni ω | |
201 | -n′ n iκ | |
202 | -#n n δ | |
203 | -#n n j | |
204 | -n′j ni ρ | |
205 | -k′ ki ω | |
206 | -k′ k iκ | |
207 | -#k k δ | |
208 | -#k k j | |
209 | -k′j ki ρ | |
210 | -g′ gi ω | |
211 | -g′ g iκ | |
212 | -#g g δ | |
213 | -g′j gi ρ | |
214 | -x ch δ | |
215 | -xʲ chi ω | |
216 | -xʲ ch iκ | |
217 | -xʲj chi ρ | |
218 | -#h h δ | |
219 | -hʲ hi ω | |
220 | -hʲ h iκ | |
221 | -mar mar z | |
222 | -m′er mier z | |
223 | -n′e nie i | |
224 | -#a a σ | |
225 | -aʲ a i | |
226 | -#a a i | |
227 | -#e e σ | |
228 | -eʲ e i | |
229 | -#e e i | |
230 | -#o o σ | |
231 | -oʲ o i | |
232 | -#o o i | |
233 | -#ó ó σ | |
234 | -óʲ ó i | |
235 | -#ó ó i | |
236 | -#u u σ | |
237 | -uʲ u i | |
238 | -#u u i | |
239 | -#i i σ | |
240 | -iʲ i i | |
241 | -#i i i | |
242 | -#y y σ | |
243 | -yʲ y i | |
244 | -#y y i | |
245 | -#ą ą σ | |
246 | -#ę ę σ | |
247 | -#j j | |
248 | -#A A | |
249 | -#B B | |
250 | -#C C | |
251 | -#Ć Ć | |
252 | -#D D | |
253 | -#E E | |
254 | -#F F | |
255 | -#G G | |
256 | -#H H | |
257 | -#I I | |
258 | -#J J | |
259 | -#K K | |
260 | -#L L | |
261 | -#Ł Ł | |
262 | -#M M | |
263 | -#N N | |
264 | -#O O | |
265 | -#P P | |
266 | -#R R | |
267 | -#S S | |
268 | -#Ś Ś | |
269 | -#T T | |
270 | -#U U | |
271 | -#W W | |
272 | -#Y Y | |
273 | -#Z Z | |
274 | -#Ż Ż | |
275 | -#- - | |
276 | -#χ́ chi | |
277 | -#h́ hi | |
269 | +#pl ř rz j | |
270 | +pl r r zi | |
271 | +#pl r r γ | |
272 | +#pl r r j | |
273 | +pl rʲ ri ω | |
274 | +pl rʲ r iκ | |
275 | +pl rj ri ρ | |
276 | +pl n′ ń λ | |
277 | +pl n′ ni ω | |
278 | +pl n′ n iκ | |
279 | +#pl n n δ | |
280 | +#pl n n j | |
281 | +pl n′j ni ρ | |
282 | +pl k′ ki ω | |
283 | +pl k′ k iκ | |
284 | +#pl k k δ | |
285 | +#pl k k j | |
286 | +pl k′j ki ρ | |
287 | +pl g′ gi ω | |
288 | +pl g′ g iκ | |
289 | +#pl g g δ | |
290 | +pl g′j gi ρ | |
291 | +pl x ch δ | |
292 | +pl xʲ chi ω | |
293 | +pl xʲ ch iκ | |
294 | +pl xʲj chi ρ | |
295 | +#pl h h δ | |
296 | +pl hʲ hi ω | |
297 | +pl hʲ h iκ | |
298 | +pl mar mar z | |
299 | +pl m′er mier z | |
300 | +pl n′e nie i | |
301 | +#pl a a σ | |
302 | +pl aʲ a i | |
303 | +#pl a a i | |
304 | +#pl e e σ | |
305 | +pl eʲ e i | |
306 | +#pl e e i | |
307 | +#pl o o σ | |
308 | +pl oʲ o i | |
309 | +#pl o o i | |
310 | +#pl ó ó σ | |
311 | +pl óʲ ó i | |
312 | +#pl ó ó i | |
313 | +#pl u u σ | |
314 | +pl uʲ u i | |
315 | +#pl u u i | |
316 | +#pl i i σ | |
317 | +pl iʲ i i | |
318 | +#pl i i i | |
319 | +#pl y y σ | |
320 | +pl yʲ y i | |
321 | +#pl y y i | |
322 | +#pl ą ą σ | |
323 | +#pl ę ę σ | |
324 | +#pl j j | |
325 | +#pl A A | |
326 | +#pl B B | |
327 | +#pl C C | |
328 | +#pl Ć Ć | |
329 | +#pl D D | |
330 | +#pl E E | |
331 | +#pl F F | |
332 | +#pl G G | |
333 | +#pl H H | |
334 | +#pl I I | |
335 | +#pl J J | |
336 | +#pl K K | |
337 | +#pl L L | |
338 | +#pl Ł Ł | |
339 | +#pl M M | |
340 | +#pl N N | |
341 | +#pl O O | |
342 | +#pl P P | |
343 | +#pl R R | |
344 | +#pl S S | |
345 | +#pl Ś Ś | |
346 | +#pl T T | |
347 | +#pl U U | |
348 | +#pl W W | |
349 | +#pl Y Y | |
350 | +#pl Z Z | |
351 | +#pl Ż Ż | |
352 | +#pl - - | |
353 | +#pl χ́ chi | |
354 | +#pl h́ hi | |
355 | + | |
356 | +de {sch}š sch | |
357 | +de {rh}r rh | |
358 | +de {z}c z | |
359 | +de {au}ał au | |
360 | +fr {q}k q | |
361 | +fr {nq}n ng | |
362 | + n n g | |
363 | +fr {qui}kʲ qui ω | |
364 | +de-fr {x}ks x | |
365 | + | |
366 | +de ks′e ksie ε | |
367 | +de {v}f′e vie ε | |
368 | +fr {v}v′i vi ε | |
369 | +fr {v}v′e vie ε | |
370 | +fr {v}v v | |
371 | + | |
372 | +de {ge}g′e ge | |
373 | +de {ke}k′e ke | |
374 | +de {ky}k′i ky | |
375 | +fr {eu}i eu chε | |
376 | +fr {eu}e eu goε | |
377 | +fr {eu}i eu mε | |
378 | +fr {eu}i eu miε | |
379 | +fr {eu}e eu muε | |
380 | +fr {li}lj li egoε | |
381 | +fr {li}lj li emuε | |
382 | +fr {li}lj li eε | |
383 | +fr {ée}e ée goε | |
384 | +fr {ée}e ée mε | |
385 | +fr {ée}e ée muε | |
386 | + | |
387 | +de {ky}kj ky ’ | |
388 | +fr {ès}e ès ’ | |
389 | +fr {ais}e ais ’ | |
390 | +fr {que}k que ’ | |
391 | +fr {phe}f phe ’ | |
392 | +fr {sy}sj sy ’ | |
393 | +fr {phy}fj phy ’ | |
394 | +fr {ry}rj ry ’ | |
395 | +fr {che}š che ’ | |
396 | +fr {de}d de ’ | |
397 | +fr {re}r re ’ | |
398 | +fr {se}z se ’ | |
399 | +fr {ge}ž ge ’ | |
400 | +fr {te}t te ’ | |
401 | +fr {le}l le ’ | |
402 | +fr {me}m me ’ | |
403 | +fr {gne}n′ gne ’ | |
404 | +fr {ne}n ne ’ | |
405 | +#fr ine ’ | |
406 | +fr {ve}v ve ’ | |
407 | +fr {gue}g gue ’ | |
408 | +fr {ce}s ce ’ | |
409 | +fr {guy}gj guy ’ | |
410 | +de-fr ’ ρ | |
411 | + | |
412 | +de góv gów ε | |
413 | +de gov′i gowi ε | |
414 | +de gov′e gowie ε | |
415 | +de gam′i gami ε | |
416 | +de gax gach ε | |
417 | +de gom gom ε | |
418 | +de kóv ków ε | |
419 | +de kov′i kowi ε | |
420 | +de kov′e kowie ε | |
421 | +de kam′i kami ε | |
422 | +de kax kach ε | |
423 | +de kom kom ε | |
424 | + | |
425 | +de k′ix kich ε | |
426 | +de k′e kie ε | |
427 | +de k′im kim ε | |
428 | +de k′im′i kimi ε | |
429 | +de k′ego kiego ε | |
430 | +de k′emu kiemu ε | |
431 | + | |
432 | +de ksax ksach ε | |
433 | +de ksa ksa ε | |
434 | +de ksam′i ksami ε | |
435 | +de ksem ksem ε | |
436 | +de ksom ksom ε | |
437 | +de ksov′e ksowie ε | |
438 | +de ksov′i ksowi ε | |
439 | +de ksóv ksów ε | |
440 | +de ksy ksy ε | |
441 | +fr ksu ksu ε | |
278 | 442 | |
443 | +de gli gli ε | |
... | ... |
morphology/fonetics.ml
... | ... | @@ -2,7 +2,7 @@ open Xstd |
2 | 2 | open Printf |
3 | 3 | |
4 | 4 | type status = Idle | Symbols | Rules | RevSymbols | RevRules |
5 | -type rule = {set: string; find: string; suf: string} | |
5 | +type rule = {set: string; find: string; suf: string; lang: string} | |
6 | 6 | |
7 | 7 | module CharTree = struct |
8 | 8 | |
... | ... | @@ -46,10 +46,10 @@ let load_rules filename = |
46 | 46 | Symbols -> status, StringMap.add symbol_defs key (Xstring.split " " vals), rev_symbol_defs, rules, rev_rules |
47 | 47 | | RevSymbols -> status, symbol_defs, StringMap.add rev_symbol_defs key (Xstring.split " " vals), rules, rev_rules |
48 | 48 | | _ -> failwith "Fonetics.load_rules: status 1") |
49 | - | [v;r;s] -> | |
49 | + | [lang;v;r;s] -> | |
50 | 50 | (match status with |
51 | - Rules -> status, symbol_defs, rev_symbol_defs, {set=v; find=r; suf=s} :: rules, rev_rules | |
52 | - | RevRules -> status, symbol_defs, rev_symbol_defs, rules, {set=r; find=v; suf=s} :: rev_rules | |
51 | + Rules -> status, symbol_defs, rev_symbol_defs, {set=v; find=r; suf=s; lang=lang} :: rules, rev_rules | |
52 | + | RevRules -> status, symbol_defs, rev_symbol_defs, rules, {set=r; find=v; suf=s; lang=lang} :: rev_rules | |
53 | 53 | | _ -> failwith "Fonetics.load_rules: status 2") |
54 | 54 | | line -> failwith ("load_rules: " ^ (String.concat "\t" line))) in |
55 | 55 | if status <> Rules && status <> RevRules then failwith "Fonetics.load_rules: status 3" else |
... | ... | @@ -66,10 +66,10 @@ let prepare_rules symbol_defs rules = |
66 | 66 | |
67 | 67 | let rules, rev_rules = |
68 | 68 | (* let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_acro.dic" in *) |
69 | - (* let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_pl.dic" in *) | |
69 | + let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_pl.dic" in | |
70 | 70 | (* let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_en.dic" in *) |
71 | 71 | (* let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_fr.dic" in *) |
72 | - let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_de.dic" in | |
72 | + (* let symbol_defs,rev_symbol_defs,rules,rev_rules = load_rules "data/fonetics_de.dic" in *) | |
73 | 73 | prepare_rules symbol_defs rules, |
74 | 74 | prepare_rules rev_symbol_defs rev_rules |
75 | 75 | |
... | ... | @@ -131,7 +131,7 @@ let rec translate_rec closure found found_maping rules s = |
131 | 131 | printf "s=%s t=%s set=%s find=%s suf=%s\n%!" s t r.set r.find r.suf); *) |
132 | 132 | let l = if l = [] && closure then |
133 | 133 | let c,s = Xunicode.first_utf8_char_of_utf8_string s in |
134 | - [s,{find=c; set=c; suf=""}] else l in | |
134 | + [s,{find=c; set=c; suf=""; lang=""}] else l in | |
135 | 135 | (* let n = String.length s in |
136 | 136 | let c = String.sub s 0 1 in |
137 | 137 | [String.sub s 1 (n-1),{find=c; set=c; suf=""}] else l in *) |
... | ... | @@ -141,7 +141,7 @@ let rec translate_rec closure found found_maping rules s = |
141 | 141 | let translate closure rules s = |
142 | 142 | (* printf "translate 1: %s\n%!" s; *) |
143 | 143 | let ll = translate_rec closure [] [] rules (s ^ "ε") in |
144 | - Xlist.rev_map ll (fun (phon,mapping) -> {Types.phon=String.concat "" phon; Types.mapping=mapping}) | |
144 | + Xlist.rev_map ll (fun (phon,mapping) -> {Types.phon=String.concat "" phon; Types.mapping=[](*mapping*)}) | |
145 | 145 | (* let s = String.concat "" l in |
146 | 146 | (* printf "translate 2: %s\n%!" s; *) |
147 | 147 | s*) |
... | ... |
morphology/generate.ml
... | ... | @@ -276,8 +276,8 @@ let _ = |
276 | 276 | find_not_validated_forms compound_rule_trees results_path verb_polimorf_filename "results/not_validated_p_verb2.tab"; *) |
277 | 277 | (* find_not_validated_forms compound_rule_trees results_path "sgjp_selected.tab" "results/not_validated_verb.tab"; *) |
278 | 278 | (* find_not_validated_forms compound_rule_trees results_path "lang_en_sgjp-20170730.tab" "results/lang_en.tab"; *) |
279 | - (* find_not_validated_forms compound_rule_trees results_path "lang_fr_sgjp-20170730.tab" "results/lang_fr.tab"; *) | |
280 | - find_not_validated_forms compound_rule_trees results_path "lang_de_sgjp-20170730.tab" "results/lang_de.tab"; | |
279 | + find_not_validated_forms compound_rule_trees results_path "lang_fr_sgjp-20170730.tab" "results/lang_fr.tab"; | |
280 | + (* find_not_validated_forms compound_rule_trees results_path "lang_de_sgjp-20170730.tab" "results/lang_de.tab"; *) | |
281 | 281 | () |
282 | 282 | |
283 | 283 | let find_not_validated_lemmata rules path filename out_filename = |
... | ... | @@ -407,6 +407,16 @@ let _ = |
407 | 407 | (* test_lemmatize "Bernoulli" "Bernoulliego"; *) |
408 | 408 | (* test_lemmatize "Handke" "Handkami"; *) |
409 | 409 | (* test_lemmatize "Radetzky" "Radetzky’ego"; *) |
410 | + (* test_lemmatize "Max" "Maksa"; *) | |
411 | + (* test_lemmatize "Montesquieu" "Montesquieugo"; *) | |
412 | + (* test_lemmatize "Java" "Javie"; *) | |
413 | + (* test_lemmatize "anglaise" "anglaise’a"; *) | |
414 | + test_lemmatize "Aristide" "Aristide’a"; | |
415 | + test_lemmatize "Beaumarchais" "Beaumarchais’go"; | |
416 | + (* test_lemmatize "Beauvoir" "Beauvoira"; *) | |
417 | + (* test_lemmatize "Bernoulli" "Bernoulliego"; *) | |
418 | + (* test_lemmatize "Astaire" "Astaire’a"; *) | |
419 | + (* test_lemmatize "Avignon" "Avignonami"; *) | |
410 | 420 | (* test_lemmatize "" ""; |
411 | 421 | test_lemmatize "" ""; |
412 | 422 | test_lemmatize "" ""; *) |
... | ... |