• 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: fa_fa_Latn_BGN.txt
6# Generated from CLDR
7#
8
9#
10########################################################################
11# BGN/PCGN 1956 System
12#
13# This system was adopted by the BGN in 1946 and by the PCGN in 1958.
14# It is used for the romanization of geographic names in Iran and
15# for Persian-language names in Afghanistan.
16#
17# Originally prepared by Michael Everson <everson@evertype.com>
18########################################################################
19#
20# MINIMAL FILTER: Persian-Latin
21#
22:: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهویي\u064E\u064F\u0650\u0651\u0652٠١٢٣٤٥٦٧٨٩پچژگی]] ;
23:: NFKD (NFC) ;
24#
25#
26########################################################################
27#
28########################################################################
29#
30# Define All Transformation Variables
31#
32########################################################################
33#
34$alef = ’;
35$ayin = ‘;
36$disambig =  \u0331 ;
37#
38#
39# Use this $wordBoundary until bug 2034 is fixed in ICU:
40# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
41#
42$wordBoundary =  [^[:L:][:M:][:N:]] ;
43#
44#
45########################################################################
46# non-letters
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
119ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
120ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
121چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
122ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
123خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
124د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
125ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
126ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
127ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
128ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
129س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
130ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
131ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
132ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
133ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
134ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
135ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
136غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
137ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
138ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
139ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
140ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
141م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
142ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
143ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
144و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
145ی\u0651 → yy ; # ARABIC LETTER FARSI YEH + SHADDA
146#
147#
148########################################################################
149#
150# End Rule 7
151#
152########################################################################
153#
154########################################################################
155#
156# Start of Transformations
157#
158########################################################################
159#
160$wordBoundary{ء →  ; # ARABIC LETTER HAMZA
161ء → $alef ; # ARABIC LETTER HAMZA
162$wordBoundary{ا →  ; # ARABIC LETTER ALEF
163آ → $alef ā ; # ARABIC FATHA ALEF WITH MADDA ABOVE
164ب → b ; # ARABIC LETTER BEH
165پ → p ; # ARABIC LETTER PEH
166ت → t ; # ARABIC LETTER TEH
167ة → h ; # ARABIC LETTER TEH MARBUTA
168ث → s\u0304 ; # ARABIC LETTER THEH
169ج → j ; # ARABIC LETTER JEEM
170چ → ch ; # ARABIC LETTER TCHEH
171ح → ḥ ; # ARABIC LETTER HAH
172خ → kh ; # ARABIC LETTER KHAH
173د → d ; # ARABIC LETTER DAL
174ذ → z\u0304 ; # ARABIC LETTER THAL
175ر → r ; # ARABIC LETTER REH
176ز → z ; # ARABIC LETTER ZAIN
177ژ → zh ; # ARABIC LETTER JEH
178س → s ; # ARABIC LETTER SEEN
179ش → sh ; # ARABIC LETTER SHEEN
180ص → ṣ ; # ARABIC LETTER SAD
181ض → ẕ ; # ARABIC LETTER DAD
182ط → ṭ ; # ARABIC LETTER TAH
183ظ → ẓ ; # ARABIC LETTER ZAH
184ع → $ayin ; # ARABIC LETTER AIN
185غ → gh ; # ARABIC LETTER GHAIN
186ف → f ; # ARABIC LETTER FEH
187ق → q ; # ARABIC LETTER QAF
188ک ↔ k ; # ARABIC LETTER KEHEH
189ك ↔ k $disambig ; # ARABIC LETTER KAF
190گ → g ; # ARABIC LETTER GAF
191ل → l ; # ARABIC LETTER LAM
192م → m ; # ARABIC LETTER MEEM
193ن → n ; # ARABIC LETTER NOON
194ه → h ; # ARABIC LETTER HEH
195و → v ; # ARABIC LETTER WAW
196ی → y ; # ARABIC LETTER FARSI YEH
197\u064Eا → ā ; # ARABIC FATHA + ALEF
198\u064Eی → á ; # ARABIC FATHA + FARSI YEH
199\u064Eو\u0652 → ow ; # ARABIC FATHA + WAW + SUKUN
200\u064E → a ; # ARABIC FATHA
201\u0650ي → ī ; # ARABIC KASRA + YEH
202\u0650 → e ; # ARABIC KASRA
203\u064Fو → ū ; # ARABIC DAMMA + WAW
204\u064F → o ; # ARABIC DAMMA
205\u0652 →  ; # ARABIC SUKUN
206::NFC (NFD) ;
207#
208#
209########################################################################
210
211