• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# © 2016 and later: Unicode, Inc. and others.
2# License & terms of use: http://www.unicode.org/copyright.html
3# Generated using tools/cldr/cldr-to-icu/build-icu-data.xml
4#
5# File: ps_ps_Latn_BGN.txt
6# Generated from CLDR
7#
8
9#
10########################################################################
11# BGN/PCGN 1968 System
12#
13# This system was adopted in 1968 for the romanization of Pashto
14# geographic names in Afghanistan. Persian names in Afghanistan are
15# romanized in accordance with the Romanization System for Persian
16# (BGN/PCGN 1958 System), shown on pages 87-92).
17#
18# Originally prepared by Michael Everson <everson@evertype.com>
19########################################################################
20#
21# MINIMAL FILTER: Pashto-Latin
22#
23:: [ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064E\u064F\u0650\u0651\u0652\u0654٠١٢٣٤٥٦٧٨٩ټپځڅچډړږژښگڰڼیۍې] ;
24:: NFD (NFC) ;
25#
26#
27########################################################################
28#
29########################################################################
30#
31# Define All Transformation Variables
32#
33########################################################################
34#
35$alef = ’;
36$ayin = ‘;
37$disambig =  \u0331 ;
38#
39#
40# Use this $wordBoundary until bug 2034 is fixed in ICU:
41# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
42#
43$wordBoundary =  [^[:L:][:M:][:N:]] ;
44#
45#
46########################################################################
47[:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
48[:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
49٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
50٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
51#  ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate
52، ↔ ',' ; # ARABIC COMMA
53؛ ↔ ';' ; # ARABIC SEMICOLON
54؟ ↔ '?' ; # ARABIC QUESTION MARK
55٪ ↔ '%' ; # ARABIC PERCENT SIGN
56٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO
57١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE
58٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO
59٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE
60٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR
61٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE
62٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX
63٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN
64٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT
65٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE
66۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO
67۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE
68۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO
69۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE
70۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR
71۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE
72۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX
73۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
74۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
75۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE
76#
77########################################################################
78#
79# Rules moved to front to avoid masking
80#
81########################################################################
82#
83########################################################################
84#
85# BGN Page 89 Rule 4
86#
87# The character sequences كه , زه , سه , and گه may be romanized k·h, z·h,
88# s·h, and g·h in order to differentiate those romanizations from the
89# digraphs kh, zh, sh, and gh.
90#
91########################################################################
92#
93كه → k·h ; # ARABIC LETTER KAF + HEH
94زه → z·h ; # ARABIC LETTER ZAIN + HEH
95سه → s·h ; # ARABIC LETTER SEEN + HEH
96گه → g·h ; # ARABIC LETTER GAF + HEH
97#
98#
99########################################################################
100#
101# End Rule 4
102#
103########################################################################
104#
105########################################################################
106#
107# BGN Page 91 Rule 7
108#
109# Doubles consonant sounds are represented in Arabic script by
110# placing a shaddah ( \u0651 ) over a consonant character. In romanization
111# the letter should be doubled. [The remainder of this rule deals with
112# the definite article and is lexical.]
113#
114########################################################################
115#
116ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA
117پ\u0651 → pp ; # ARABIC LETTER PEH + SHADDA
118ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
119ټ\u0651 → ṯṯ ; # ARABIC LETTER TEH WITH RING + SHADDA
120ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
121ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
122چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
123\u0651څ → tsts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE + SHADDA
124\u0651ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE + SHADDA
125ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
126خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
127د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
128\u0651ډ → ḏḏ ; # ARABIC LETTER DAL WITH RING + SHADDA
129ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
130ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
131\u0651ړ → ṟṟ ; # ARABIC LETTER REH WITH RING + SHADDA
132ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
133ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
134\u0651ژ → z\u035Fhz\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE + SHADDA
135س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
136ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
137\u0651ښ → s\u035Fhs\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
138ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
139ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
140ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
141ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
142ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
143غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
144ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
145ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
146ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
147\u0651گ → gg ; # ARABIC LETTER GAF + SHADDA
148\u0651ڰ → gg ; # ARABIC LETTER GAF WITH RING + SHADDA
149ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
150م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
151ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
152\u0651ڼ → ṉṉ ; # ARABIC LETTER NOON WITH RING + SHADDA
153ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
154و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
155\u0651ی → yy ; # ARABIC LETTER FARSI YEH + SHADDA
156ى\u0651 → yy ; # ARABIC LETTER YEH + SHADDA
157#
158#
159########################################################################
160#
161# End Rule 7
162#
163########################################################################
164#
165########################################################################
166#
167# Start of Transformations
168#
169########################################################################
170#
171$wordBoundary{ء →  ; # ARABIC LETTER HAMZA
172ء → $alef ; # ARABIC LETTER HAMZA
173$wordBoundary{ا →  ; # ARABIC LETTER ALEF
174آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
175ب → b ; # ARABIC LETTER BEH
176پ → p ; # ARABIC LETTER PEH
177ت → t ; # ARABIC LETTER TEH
178ټ → ṯ ; # ARABIC LETTER TEH WITH RING
179ة → h ; # ARABIC LETTER TEH MARBUTA
180ث → s\u0304 ; # ARABIC LETTER THEH
181ج → j ; # ARABIC LETTER JEEM
182چ → ch ; # ARABIC LETTER TCHEH
183څ → ts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE
184ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE
185ح → ḥ ; # ARABIC LETTER HAH
186خ → kh ; # ARABIC LETTER KHAH
187د → d ; # ARABIC LETTER DAL
188ډ → ḏ ; # ARABIC LETTER DAL WITH RING
189ذ → z\u0304 ; # ARABIC LETTER THAL
190ر → r ; # ARABIC LETTER REH
191ړ → ṟ ; # ARABIC LETTER REH WITH RING
192ز → z ; # ARABIC LETTER ZAIN
193ژ → zh ; # ARABIC LETTER JEH
194ږ → z\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE
195س → s ; # ARABIC LETTER SEEN
196ش → sh ; # ARABIC LETTER SHEEN
197ښ → s\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
198ص → ṣ ; # ARABIC LETTER SAD
199ض → ẕ ; # ARABIC LETTER DAD
200ط → ṭ ; # ARABIC LETTER TAH
201ظ → ẓ ; # ARABIC LETTER ZAH
202ع → $ayin ; # ARABIC LETTER AIN
203غ → gh ; # ARABIC LETTER GHAIN
204ف → f ; # ARABIC LETTER FEH
205ق → q ; # ARABIC LETTER QAF
206ك → k ; # ARABIC LETTER KAF
207گ → g ; # ARABIC LETTER GAF
208ڰ → g ; # ARABIC LETTER GAF WITH RING
209ل → l ; # ARABIC LETTER LAM
210م → m ; # ARABIC LETTER MEEM
211ن → n ; # ARABIC LETTER NOON
212ڼ → ṉ ; # ARABIC LETTER NOON WITH RING
213و → w ; # ARABIC LETTER WAW
214ه → h ; # ARABIC LETTER HEH
215\u0654ی → ey ; # ARABIC LETTER FARSI YEH + HAMZA ABOVE
216ی → y ; # ARABIC LETTER FARSI YEH
217ى → y ; # ARABIC LETTER YEH
218ې → e ; # ARABIC LETTER E
219\u064Eا → ā ; # ARABIC FATHA + ALEF
220\u064Eى\u0652 → ay ; # ARABIC FATHA + FARSI YEH + SUKUN
221\u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA
222\u064E\u0652ۍ → êy ; # ARABIC FATHA + SUKUN + YEH WITH TAIL
223\u064E\u0652 → ê ; # ARABIC FATHA + SUKUN
224\u064E → a ; # ARABIC FATHA
225\u0650\u0652ي → ey ; # ARABIC KASRA + FARSI YEH + SUKUN
226\u0650ي → ī ; # ARABIC KASRA + FARSI YEH
227\u0650ى → ī ; # ARABIC KASRA + YEH
228\u0650و → ew ; # ARABIC KASRA + WAW
229\u0650 → i ; # ARABIC KASRA
230\u064Fو\u0652 → ow ; # ARABIC DAMMA + WAW + SUKUN
231\u064Fوی → ūy ; # ARABIC DAMMA + WAW + FARSI YEH
232\u064Fو → ū ; # ARABIC DAMMA + WAW
233\u064F → u ; # ARABIC DAMMA
234\u0652 →  ; # ARABIC SUKUN
235#
236#
237########################################################################
238
239