Line 0
Link Here
|
0 |
- |
1 |
# Generic character map for Latin and Greek characters. |
|
|
2 |
|
3 |
# Define the basic value-set. *Beware* of changing this without re-indexing |
4 |
# your databases. |
5 |
encoding utf-8 |
6 |
|
7 |
# basic character set |
8 |
lowercase {0-9}{a-z}αβγδεζηθικλμνξοπρστυφχψω |
9 |
uppercase {0-9}{A-Z}ΑΒΓΔΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩ |
10 |
|
11 |
# Breaking characters |
12 |
space {\001-\040}!"#$%&'\()*+,-./:;<=>?@\[\\]^_`\{|}~’{\x88-\x89}{\x98-\x9C}¡¿«» |
13 |
|
14 |
# Characters to be considered equivalent |
15 |
equivalent aáàãåâăąȧǎȁȃ |
16 |
equivalent Æӕä(ae) |
17 |
equivalent ā(aa) |
18 |
equivalent iíìîịĩĭįǐȉȋ |
19 |
equivalent ï(ie) |
20 |
equivalent ī(ii) |
21 |
equivalent uúùûũŭųűǔȕȗ |
22 |
equivalent ü(ue) |
23 |
equivalent ū(uu) |
24 |
equivalent eéèêẽĕęėěȅȇ |
25 |
equivalent ëē(ee) |
26 |
equivalent oóòõôŏǫȯőǒȍȏ |
27 |
equivalent Œœöø(oe) |
28 |
equivalent ō(oo) |
29 |
equivalent αάΑΆ(α΄)(Α΄) |
30 |
equivalent εέΕΈ(ε΄)(Ε΄) |
31 |
equivalent ιίϊΙΊΪ(ι΄)(Ι΄) |
32 |
equivalent ηήΗΉ(η΄)(Η΄) |
33 |
equivalent οόΟΌ(ο΄)(Ο΄) |
34 |
equivalent υύϋΥΎΫ(υ΄)(Υ΄) |
35 |
equivalent ωώΩΏ(ω΄)(Ω΄) |
36 |
equivalent Σσς |
37 |
|
38 |
# Supplemental mappings |
39 |
# mapping diacritics to plain ASCII |
40 |
# A |
41 |
map ā a |
42 |
map Ā a |
43 |
map á a |
44 |
map Á a |
45 |
map à a |
46 |
map À a |
47 |
map ã a |
48 |
map à a |
49 |
map å a |
50 |
map Å a |
51 |
map â a |
52 |
map  a |
53 |
map ä a |
54 |
map Ä a |
55 |
map ă a |
56 |
map Ă a |
57 |
map ą a |
58 |
map Ą a |
59 |
map ȧ a |
60 |
map Ȧ a |
61 |
map ǎ a |
62 |
map Ǎ a |
63 |
map ȁ a |
64 |
map Ȁ a |
65 |
map ȃ a |
66 |
map Ȃ a |
67 |
map ⱥ a |
68 |
map Ⱥ a |
69 |
# AE |
70 |
map ӕ (ae) |
71 |
map Ӕ (ae) |
72 |
# I |
73 |
map ī i |
74 |
map Ī i |
75 |
map ï i |
76 |
map Ï i |
77 |
map î i |
78 |
map Î i |
79 |
map í i |
80 |
map Í i |
81 |
map ì i |
82 |
map Ì i |
83 |
map ị i |
84 |
map Ị i |
85 |
map ĩ i |
86 |
map Ĩ i |
87 |
map ĭ i |
88 |
map Ĭ i |
89 |
map į i |
90 |
map Į i |
91 |
map ǐ i |
92 |
map Ǐ i |
93 |
map ȉ i |
94 |
map Ȉ i |
95 |
map ȋ i |
96 |
map Ȋ i |
97 |
map ɨ i |
98 |
map Ɨ i |
99 |
# U |
100 |
map ū u |
101 |
map Ū u |
102 |
map ü u |
103 |
map Ü u |
104 |
map ù u |
105 |
map Ù u |
106 |
map ú u |
107 |
map Ú u |
108 |
map û u |
109 |
map Û u |
110 |
map ũ u |
111 |
map Ũ u |
112 |
map ŭ u |
113 |
map Ŭ u |
114 |
map ų u |
115 |
map Ų u |
116 |
map ű u |
117 |
map Ű u |
118 |
map ǔ u |
119 |
map Ǔ u |
120 |
map ȕ u |
121 |
map Ȕ u |
122 |
map ȗ u |
123 |
map Ȗ u |
124 |
map ʉ u |
125 |
map Ʉ u |
126 |
map ů u |
127 |
map Ů u |
128 |
# E |
129 |
map ê e |
130 |
map Ê e |
131 |
map ë e |
132 |
map Ë e |
133 |
map é e |
134 |
map É e |
135 |
map è e |
136 |
map È e |
137 |
map ē e |
138 |
map Ē e |
139 |
map ẽ e |
140 |
map Ẽ e |
141 |
map ĕ e |
142 |
map Ĕ e |
143 |
map ę e |
144 |
map Ę e |
145 |
map ė e |
146 |
map Ė e |
147 |
map ě e |
148 |
map Ě e |
149 |
map ȅ e |
150 |
map Ȅ e |
151 |
map ȇ e |
152 |
map Ȇ e |
153 |
map ɇ e |
154 |
map Ɇ e |
155 |
# O |
156 |
map ô o |
157 |
map Ô o |
158 |
map ó o |
159 |
map Ó o |
160 |
map ò o |
161 |
map Ò o |
162 |
map ö o |
163 |
map Ö o |
164 |
map õ o |
165 |
map Õ o |
166 |
map ø o |
167 |
map Ø o |
168 |
map ō o |
169 |
map Ō o |
170 |
map ǒ o |
171 |
map Ǒ o |
172 |
map ŏ o |
173 |
map Ŏ o |
174 |
map ǫ o |
175 |
map Ǫ o |
176 |
map ȯ o |
177 |
map Ȯ o |
178 |
map ő o |
179 |
map Ő o |
180 |
map ȍ o |
181 |
map Ȍ o |
182 |
map ȏ o |
183 |
map Ȏ o |
184 |
map Œ (oe) |
185 |
map œ (oe) |
186 |
# R |
187 |
map ṛ r |
188 |
map Ṛ r |
189 |
map r̥ r |
190 |
map R̥ r |
191 |
map ṝ r |
192 |
map Ṝ r |
193 |
map ř r |
194 |
map Ř r |
195 |
map ṟ r |
196 |
map ɍ r |
197 |
map Ɍ r |
198 |
# L |
199 |
map ḷ l |
200 |
map Ḷ l |
201 |
map ḹ l |
202 |
map Ḹ l |
203 |
map ḻ l |
204 |
map Ḻ l |
205 |
map ł l |
206 |
map Ł l |
207 |
# N |
208 |
map ṅ n |
209 |
map Ṅ n |
210 |
map ñ n |
211 |
map Ñ n |
212 |
map ǹ n |
213 |
map Ǹ n |
214 |
map ń n |
215 |
map Ń n |
216 |
map ṉ n |
217 |
map Ṉ n |
218 |
map ṇ n |
219 |
map Ṇ n |
220 |
# T |
221 |
map ṭ t |
222 |
map Ṭ t |
223 |
map ŧ t |
224 |
map Ŧ t |
225 |
# D |
226 |
map ḍ d |
227 |
map Ḍ d |
228 |
map đ d |
229 |
map Đ d |
230 |
# S |
231 |
map ś s |
232 |
map Ś s |
233 |
map ṣ s |
234 |
map Ṣ s |
235 |
map ş s |
236 |
map Ş s |
237 |
map š s |
238 |
map Š s |
239 |
map ṡ s |
240 |
map Ṡ s |
241 |
# M |
242 |
map ṁ m |
243 |
map Ṁ m |
244 |
map ṃ m |
245 |
map Ṃ m |
246 |
# H |
247 |
map ḥ h |
248 |
map Ḥ h |
249 |
map ẖ h |
250 |
map ħ h |
251 |
map Ħ h |
252 |
# Z |
253 |
map ź z |
254 |
map Ź z |
255 |
map ẓ z |
256 |
map Ẓ z |
257 |
map ż z |
258 |
map Ż z |
259 |
map ƶ z |
260 |
map Ƶ z |
261 |
#C |
262 |
map ç c |
263 |
map Ç c |
264 |
map ć c |
265 |
map Ć c |
266 |
map č c |
267 |
map Č c |
268 |
map ȼ c |
269 |
map Ȼ c |
270 |
map Ċ c |
271 |
map ċ c |
272 |
map Ĉ c |
273 |
map ĉ c |
274 |
# G |
275 |
map ǥ g |
276 |
map Ǥ g |
277 |
map ġ g |
278 |
map Ġ g |
279 |
map ǧ g |
280 |
map Ǧ g |
281 |
map ğ g |
282 |
map Ğ g |
283 |
# K |
284 |
map ḵ k |
285 |
map Ḵ k |
286 |
map Ḱ k |
287 |
map ḱ k |
288 |
# Y |
289 |
map ý y |
290 |
map Ý y |
291 |
map ɏ y |
292 |
map Ɏ y |
293 |
|
294 |
## mapping diacritics to plain on greek characters |
295 |
map ά α |
296 |
map Ά α |
297 |
map έ ε |
298 |
map Έ ε |
299 |
map ί ι |
300 |
map Ί ι |
301 |
map ή η |
302 |
map Ή η |
303 |
map ό ο |
304 |
map Ό ο |
305 |
map ύ υ |
306 |
map Ύ υ |
307 |
map ώ ω |
308 |
map Ώ ω |
309 |
map Σ σ |
310 |
map ς σ |
311 |
|
312 |
# Remove prefixes |
313 |
map (^ὁ\s) @ |
314 |
map (^ἡ\s) @ |
315 |
map (^τό\s) @ |
316 |
map (^τώ\s) @ |
317 |
map (^οἱ\s) @ |
318 |
map (^αἱ\s) @ |
319 |
map (^τά\s) @ |