]>
Commit | Line | Data |
---|---|---|
f3c0d7a5 A |
1 | # © 2016 and later: Unicode, Inc. and others. |
2 | # License & terms of use: http://www.unicode.org/copyright.html#License | |
3 | # | |
2ca993e8 | 4 | # File: ps_ps_Latn_BGN.txt |
f3c0d7a5 | 5 | # Generated from CLDR |
729e4ab9 | 6 | # |
2ca993e8 A |
7 | |
8 | # | |
9 | ######################################################################## | |
10 | # BGN/PCGN 1968 System | |
11 | # | |
12 | # This system was adopted in 1968 for the romanization of Pashto | |
13 | # geographic names in Afghanistan. Persian names in Afghanistan are | |
14 | # romanized in accordance with the Romanization System for Persian | |
15 | # (BGN/PCGN 1958 System), shown on pages 87-92). | |
16 | # | |
17 | # Originally prepared by Michael Everson <everson@evertype.com> | |
18 | ######################################################################## | |
19 | # | |
20 | # MINIMAL FILTER: Pashto-Latin | |
21 | # | |
729e4ab9 A |
22 | :: [ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064E\u064F\u0650\u0651\u0652\u0654٠١٢٣٤٥٦٧٨٩ټپځڅچډړږژښگڰڼیۍې] ; |
23 | :: NFD (NFC) ; | |
2ca993e8 A |
24 | # |
25 | # | |
26 | ######################################################################## | |
27 | # | |
28 | ######################################################################## | |
29 | # | |
30 | # Define All Transformation Variables | |
31 | # | |
32 | ######################################################################## | |
33 | # | |
729e4ab9 A |
34 | $alef = ’; |
35 | $ayin = ‘; | |
51004dcb | 36 | $disambig = \u0331 ; |
2ca993e8 A |
37 | # |
38 | # | |
39 | # Use this $wordBoundary until bug 2034 is fixed in ICU: | |
40 | # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest | |
41 | # | |
51004dcb | 42 | $wordBoundary = [^[:L:][:M:][:N:]] ; |
2ca993e8 A |
43 | # |
44 | # | |
45 | ######################################################################## | |
729e4ab9 A |
46 | [:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR |
47 | [:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR | |
48 | ٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR | |
49 | ٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR | |
2ca993e8 | 50 | # ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate |
729e4ab9 A |
51 | ، ↔ ',' ; # ARABIC COMMA |
52 | ؛ ↔ ';' ; # ARABIC SEMICOLON | |
53 | ؟ ↔ '?' ; # ARABIC QUESTION MARK | |
54 | ٪ ↔ '%' ; # ARABIC PERCENT SIGN | |
55 | ٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO | |
56 | ١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE | |
57 | ٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO | |
58 | ٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE | |
59 | ٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR | |
60 | ٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE | |
61 | ٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX | |
62 | ٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN | |
63 | ٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT | |
64 | ٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE | |
65 | ۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO | |
66 | ۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE | |
67 | ۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO | |
68 | ۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE | |
69 | ۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR | |
70 | ۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE | |
71 | ۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX | |
72 | ۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN | |
73 | ۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT | |
74 | ۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE | |
2ca993e8 A |
75 | # |
76 | ######################################################################## | |
77 | # | |
78 | # Rules moved to front to avoid masking | |
79 | # | |
80 | ######################################################################## | |
81 | # | |
82 | ######################################################################## | |
83 | # | |
84 | # BGN Page 89 Rule 4 | |
85 | # | |
86 | # The character sequences كه , زه , سه , and گه may be romanized k·h, z·h, | |
87 | # s·h, and g·h in order to differentiate those romanizations from the | |
88 | # digraphs kh, zh, sh, and gh. | |
89 | # | |
90 | ######################################################################## | |
91 | # | |
729e4ab9 A |
92 | كه → k·h ; # ARABIC LETTER KAF + HEH |
93 | زه → z·h ; # ARABIC LETTER ZAIN + HEH | |
94 | سه → s·h ; # ARABIC LETTER SEEN + HEH | |
95 | گه → g·h ; # ARABIC LETTER GAF + HEH | |
2ca993e8 A |
96 | # |
97 | # | |
98 | ######################################################################## | |
99 | # | |
100 | # End Rule 4 | |
101 | # | |
102 | ######################################################################## | |
103 | # | |
104 | ######################################################################## | |
105 | # | |
106 | # BGN Page 91 Rule 7 | |
107 | # | |
108 | # Doubles consonant sounds are represented in Arabic script by | |
109 | # placing a shaddah ( \u0651 ) over a consonant character. In romanization | |
110 | # the letter should be doubled. [The remainder of this rule deals with | |
111 | # the definite article and is lexical.] | |
112 | # | |
113 | ######################################################################## | |
114 | # | |
729e4ab9 A |
115 | ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA |
116 | پ\u0651 → pp ; # ARABIC LETTER PEH + SHADDA | |
117 | ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA | |
118 | ټ\u0651 → ṯṯ ; # ARABIC LETTER TEH WITH RING + SHADDA | |
119 | ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA | |
120 | ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA | |
121 | چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA | |
122 | \u0651څ → tsts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE + SHADDA | |
123 | \u0651ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE + SHADDA | |
124 | ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA | |
125 | خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA | |
126 | د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA | |
127 | \u0651ډ → ḏḏ ; # ARABIC LETTER DAL WITH RING + SHADDA | |
128 | ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA | |
129 | ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA | |
130 | \u0651ړ → ṟṟ ; # ARABIC LETTER REH WITH RING + SHADDA | |
131 | ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA | |
132 | ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA | |
133 | \u0651ژ → z\u035Fhz\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE + SHADDA | |
134 | س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA | |
135 | ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA | |
136 | \u0651ښ → s\u035Fhs\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE | |
137 | ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA | |
138 | ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA | |
139 | ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA | |
140 | ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA | |
141 | ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA | |
142 | غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA | |
143 | ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA | |
144 | ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA | |
145 | ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA | |
146 | \u0651گ → gg ; # ARABIC LETTER GAF + SHADDA | |
147 | \u0651ڰ → gg ; # ARABIC LETTER GAF WITH RING + SHADDA | |
148 | ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA | |
149 | م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA | |
150 | ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA | |
151 | \u0651ڼ → ṉṉ ; # ARABIC LETTER NOON WITH RING + SHADDA | |
152 | ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA | |
153 | و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA | |
154 | \u0651ی → yy ; # ARABIC LETTER FARSI YEH + SHADDA | |
155 | ى\u0651 → yy ; # ARABIC LETTER YEH + SHADDA | |
2ca993e8 A |
156 | # |
157 | # | |
158 | ######################################################################## | |
159 | # | |
160 | # End Rule 7 | |
161 | # | |
162 | ######################################################################## | |
163 | # | |
164 | ######################################################################## | |
165 | # | |
166 | # Start of Transformations | |
167 | # | |
168 | ######################################################################## | |
169 | # | |
51004dcb | 170 | $wordBoundary{ء → ; # ARABIC LETTER HAMZA |
729e4ab9 | 171 | ء → $alef ; # ARABIC LETTER HAMZA |
51004dcb | 172 | $wordBoundary{ا → ; # ARABIC LETTER ALEF |
729e4ab9 A |
173 | آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE |
174 | ب → b ; # ARABIC LETTER BEH | |
175 | پ → p ; # ARABIC LETTER PEH | |
176 | ت → t ; # ARABIC LETTER TEH | |
177 | ټ → ṯ ; # ARABIC LETTER TEH WITH RING | |
178 | ة → h ; # ARABIC LETTER TEH MARBUTA | |
179 | ث → s\u0304 ; # ARABIC LETTER THEH | |
180 | ج → j ; # ARABIC LETTER JEEM | |
181 | چ → ch ; # ARABIC LETTER TCHEH | |
182 | څ → ts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE | |
183 | ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE | |
184 | ح → ḥ ; # ARABIC LETTER HAH | |
185 | خ → kh ; # ARABIC LETTER KHAH | |
186 | د → d ; # ARABIC LETTER DAL | |
187 | ډ → ḏ ; # ARABIC LETTER DAL WITH RING | |
188 | ذ → z\u0304 ; # ARABIC LETTER THAL | |
189 | ر → r ; # ARABIC LETTER REH | |
190 | ړ → ṟ ; # ARABIC LETTER REH WITH RING | |
191 | ز → z ; # ARABIC LETTER ZAIN | |
192 | ژ → zh ; # ARABIC LETTER JEH | |
193 | ږ → z\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE | |
194 | س → s ; # ARABIC LETTER SEEN | |
195 | ش → sh ; # ARABIC LETTER SHEEN | |
196 | ښ → s\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE | |
197 | ص → ṣ ; # ARABIC LETTER SAD | |
198 | ض → ẕ ; # ARABIC LETTER DAD | |
199 | ط → ṭ ; # ARABIC LETTER TAH | |
200 | ظ → ẓ ; # ARABIC LETTER ZAH | |
201 | ع → $ayin ; # ARABIC LETTER AIN | |
202 | غ → gh ; # ARABIC LETTER GHAIN | |
203 | ف → f ; # ARABIC LETTER FEH | |
204 | ق → q ; # ARABIC LETTER QAF | |
205 | ك → k ; # ARABIC LETTER KAF | |
206 | گ → g ; # ARABIC LETTER GAF | |
207 | ڰ → g ; # ARABIC LETTER GAF WITH RING | |
208 | ل → l ; # ARABIC LETTER LAM | |
209 | م → m ; # ARABIC LETTER MEEM | |
210 | ن → n ; # ARABIC LETTER NOON | |
211 | ڼ → ṉ ; # ARABIC LETTER NOON WITH RING | |
212 | و → w ; # ARABIC LETTER WAW | |
213 | ه → h ; # ARABIC LETTER HEH | |
214 | \u0654ی → ey ; # ARABIC LETTER FARSI YEH + HAMZA ABOVE | |
215 | ی → y ; # ARABIC LETTER FARSI YEH | |
216 | ى → y ; # ARABIC LETTER YEH | |
217 | ې → e ; # ARABIC LETTER E | |
218 | \u064Eا → ā ; # ARABIC FATHA + ALEF | |
2ca993e8 | 219 | \u064Eى\u0652 → ay ; # ARABIC FATHA + FARSI YEH + SUKUN |
729e4ab9 A |
220 | \u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA |
221 | \u064E\u0652ۍ → êy ; # ARABIC FATHA + SUKUN + YEH WITH TAIL | |
222 | \u064E\u0652 → ê ; # ARABIC FATHA + SUKUN | |
223 | \u064E → a ; # ARABIC FATHA | |
224 | \u0650\u0652ي → ey ; # ARABIC KASRA + FARSI YEH + SUKUN | |
225 | \u0650ي → ī ; # ARABIC KASRA + FARSI YEH | |
226 | \u0650ى → ī ; # ARABIC KASRA + YEH | |
227 | \u0650و → ew ; # ARABIC KASRA + WAW | |
228 | \u0650 → i ; # ARABIC KASRA | |
229 | \u064Fو\u0652 → ow ; # ARABIC DAMMA + WAW + SUKUN | |
230 | \u064Fوی → ūy ; # ARABIC DAMMA + WAW + FARSI YEH | |
231 | \u064Fو → ū ; # ARABIC DAMMA + WAW | |
232 | \u064F → u ; # ARABIC DAMMA | |
51004dcb | 233 | \u0652 → ; # ARABIC SUKUN |
2ca993e8 A |
234 | # |
235 | # | |
236 | ######################################################################## | |
237 |