[apple/icu.git] / icuSources / data / translit / fa_fa_Latn_BGN.txt

# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html#License
#
# File: fa_fa_Latn_BGN.txt
# Generated from CLDR
#

#
########################################################################
# BGN/PCGN 1956 System
#
# This system was adopted by the BGN in 1946 and by the PCGN in 1958.
# It is used for the romanization of geographic names in Iran and
# for Persian-language names in Afghanistan.
#
# Originally prepared by Michael Everson <everson@evertype.com>
########################################################################
#
# MINIMAL FILTER: Persian-Latin
#
:: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهویي\u064E\u064F\u0650\u0651\u0652٠١٢٣٤٥٦٧٨٩پچژگی]] ;
:: NFKD (NFC) ;
#
#
########################################################################
#
########################################################################
#
# Define All Transformation Variables
#
########################################################################
#
$alef = ’;
$ayin = ‘;
$disambig =  \u0331 ;
#
#
# Use this $wordBoundary until bug 2034 is fixed in ICU:
# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
#
$wordBoundary =  [^[:L:][:M:][:N:]] ;
#
#
########################################################################
# non-letters
[:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
[:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
#  ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate
، ↔ ',' ; # ARABIC COMMA
؛ ↔ ';' ; # ARABIC SEMICOLON
؟ ↔ '?' ; # ARABIC QUESTION MARK
٪ ↔ '%' ; # ARABIC PERCENT SIGN
٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO
١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE
٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO
٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE
٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR
٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE
٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX
٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN
٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT
٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE
۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO
۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE
۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO
۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE
۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR
۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE
۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX
۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE
#
########################################################################
#
# Rules moved to front to avoid masking
#
########################################################################
#
########################################################################
#
# BGN Page 89 Rule 4
#
# The character sequences كه , زه , سه , and گه may be romanized k·h, z·h,
# s·h, and g·h in order to differentiate those romanizations from the
# digraphs kh, zh, sh, and gh.
#
########################################################################
#
كه → k·h ; # ARABIC LETTER KAF + HEH
زه → z·h ; # ARABIC LETTER ZAIN + HEH
سه → s·h ; # ARABIC LETTER SEEN + HEH
گه → g·h ; # ARABIC LETTER GAF + HEH
#
#
########################################################################
#
# End Rule 4
#
########################################################################
#
########################################################################
#
# BGN Page 91 Rule 7
#
# Doubles consonant sounds are represented in Arabic script by
# placing a shaddah ( \u0651 ) over a consonant character. In romanization
# the letter should be doubled. [The remainder of this rule deals with
# the definite article and is lexical.]
#
########################################################################
#
ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA
پ\u0651 → pp ; # ARABIC LETTER PEH + SHADDA
ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
ی\u0651 → yy ; # ARABIC LETTER FARSI YEH + SHADDA
#
#
########################################################################
#
# End Rule 7
#
########################################################################
#
########################################################################
#
# Start of Transformations
#
########################################################################
#
$wordBoundary{ء →  ; # ARABIC LETTER HAMZA
ء → $alef ; # ARABIC LETTER HAMZA
$wordBoundary{ا →  ; # ARABIC LETTER ALEF
آ → $alef ā ; # ARABIC FATHA ALEF WITH MADDA ABOVE
ب → b ; # ARABIC LETTER BEH
پ → p ; # ARABIC LETTER PEH
ت → t ; # ARABIC LETTER TEH
ة → h ; # ARABIC LETTER TEH MARBUTA
ث → s\u0304 ; # ARABIC LETTER THEH
ج → j ; # ARABIC LETTER JEEM
چ → ch ; # ARABIC LETTER TCHEH
ح → ḥ ; # ARABIC LETTER HAH
خ → kh ; # ARABIC LETTER KHAH
د → d ; # ARABIC LETTER DAL
ذ → z\u0304 ; # ARABIC LETTER THAL
ر → r ; # ARABIC LETTER REH
ز → z ; # ARABIC LETTER ZAIN
ژ → zh ; # ARABIC LETTER JEH
س → s ; # ARABIC LETTER SEEN
ش → sh ; # ARABIC LETTER SHEEN
ص → ṣ ; # ARABIC LETTER SAD
ض → ẕ ; # ARABIC LETTER DAD
ط → ṭ ; # ARABIC LETTER TAH
ظ → ẓ ; # ARABIC LETTER ZAH
ع → $ayin ; # ARABIC LETTER AIN
غ → gh ; # ARABIC LETTER GHAIN
ف → f ; # ARABIC LETTER FEH
ق → q ; # ARABIC LETTER QAF
ک ↔ k ; # ARABIC LETTER KEHEH
ك ↔ k $disambig ; # ARABIC LETTER KAF
گ → g ; # ARABIC LETTER GAF
ل → l ; # ARABIC LETTER LAM
م → m ; # ARABIC LETTER MEEM
ن → n ; # ARABIC LETTER NOON
ه → h ; # ARABIC LETTER HEH
و → v ; # ARABIC LETTER WAW
ی → y ; # ARABIC LETTER FARSI YEH
\u064Eا → ā ; # ARABIC FATHA + ALEF
\u064Eی → á ; # ARABIC FATHA + FARSI YEH
\u064Eو\u0652 → ow ; # ARABIC FATHA + WAW + SUKUN
\u064E → a ; # ARABIC FATHA
\u0650ي → ī ; # ARABIC KASRA + YEH
\u0650 → e ; # ARABIC KASRA
\u064Fو → ū ; # ARABIC DAMMA + WAW
\u064F → o ; # ARABIC DAMMA
\u0652 →  ; # ARABIC SUKUN
::NFC (NFD) ;
#
#
########################################################################
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: fa_fa_Latn_BGN.txt
f3c0d7a5	5	# Generated from CLDR
729e4ab9	6	#
2ca993e8 A	7
	8	#
	9	########################################################################
	10	# BGN/PCGN 1956 System
	11	#
	12	# This system was adopted by the BGN in 1946 and by the PCGN in 1958.
	13	# It is used for the romanization of geographic names in Iran and
	14	# for Persian-language names in Afghanistan.
	15	#
	16	# Originally prepared by Michael Everson <everson@evertype.com>
	17	########################################################################
	18	#
	19	# MINIMAL FILTER: Persian-Latin
	20	#
729e4ab9 A	21	:: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهویي\u064E\u064F\u0650\u0651\u0652٠١٢٣٤٥٦٧٨٩پچژگی]] ;
729e4ab9 A	22	:: NFKD (NFC) ;
2ca993e8 A	23	#
	24	#
	25	########################################################################
	26	#
	27	########################################################################
	28	#
	29	# Define All Transformation Variables
	30	#
	31	########################################################################
	32	#
729e4ab9 A	33	$alef = ’;
729e4ab9 A	34	$ayin = ‘;
51004dcb	35	$disambig = \u0331 ;
2ca993e8 A	36	#
	37	#
	38	# Use this $wordBoundary until bug 2034 is fixed in ICU:
	39	# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
	40	#
51004dcb	41	$wordBoundary = [^[:L:][:M:][:N:]] ;
2ca993e8 A	42	#
	43	#
	44	########################################################################
	45	# non-letters
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	ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
	119	ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
	120	چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
	121	ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
	122	خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
	123	د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
	124	ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
	125	ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
	126	ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
	127	ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
	128	س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
	129	ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
	130	ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
	131	ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
	132	ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
	133	ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
	134	ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
	135	غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
	136	ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
	137	ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
	138	ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
	139	ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
	140	م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
	141	ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
	142	ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
	143	و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
	144	ی\u0651 → yy ; # ARABIC LETTER FARSI YEH + SHADDA
2ca993e8 A	145	#
	146	#
	147	########################################################################
	148	#
	149	# End Rule 7
	150	#
	151	########################################################################
	152	#
	153	########################################################################
	154	#
	155	# Start of Transformations
	156	#
	157	########################################################################
	158	#
51004dcb	159	$wordBoundary{ء → ; # ARABIC LETTER HAMZA
729e4ab9	160	ء → $alef ; # ARABIC LETTER HAMZA
51004dcb	161	$wordBoundary{ا → ; # ARABIC LETTER ALEF
729e4ab9 A	162	آ → $alef ā ; # ARABIC FATHA ALEF WITH MADDA ABOVE
	163	ب → b ; # ARABIC LETTER BEH
	164	پ → p ; # ARABIC LETTER PEH
	165	ت → t ; # ARABIC LETTER TEH
	166	ة → h ; # ARABIC LETTER TEH MARBUTA
	167	ث → s\u0304 ; # ARABIC LETTER THEH
	168	ج → j ; # ARABIC LETTER JEEM
	169	چ → ch ; # ARABIC LETTER TCHEH
	170	ح → ḥ ; # ARABIC LETTER HAH
	171	خ → kh ; # ARABIC LETTER KHAH
	172	د → d ; # ARABIC LETTER DAL
	173	ذ → z\u0304 ; # ARABIC LETTER THAL
	174	ر → r ; # ARABIC LETTER REH
	175	ز → z ; # ARABIC LETTER ZAIN
	176	ژ → zh ; # ARABIC LETTER JEH
	177	س → s ; # ARABIC LETTER SEEN
	178	ش → sh ; # ARABIC LETTER SHEEN
	179	ص → ṣ ; # ARABIC LETTER SAD
	180	ض → ẕ ; # ARABIC LETTER DAD
	181	ط → ṭ ; # ARABIC LETTER TAH
	182	ظ → ẓ ; # ARABIC LETTER ZAH
	183	ع → $ayin ; # ARABIC LETTER AIN
	184	غ → gh ; # ARABIC LETTER GHAIN
	185	ف → f ; # ARABIC LETTER FEH
	186	ق → q ; # ARABIC LETTER QAF
	187	ک ↔ k ; # ARABIC LETTER KEHEH
	188	ك ↔ k $disambig ; # ARABIC LETTER KAF
	189	گ → g ; # ARABIC LETTER GAF
	190	ل → l ; # ARABIC LETTER LAM
	191	م → m ; # ARABIC LETTER MEEM
	192	ن → n ; # ARABIC LETTER NOON
	193	ه → h ; # ARABIC LETTER HEH
	194	و → v ; # ARABIC LETTER WAW
	195	ی → y ; # ARABIC LETTER FARSI YEH
	196	\u064Eا → ā ; # ARABIC FATHA + ALEF
	197	\u064Eی → á ; # ARABIC FATHA + FARSI YEH
	198	\u064Eو\u0652 → ow ; # ARABIC FATHA + WAW + SUKUN
	199	\u064E → a ; # ARABIC FATHA
	200	\u0650ي → ī ; # ARABIC KASRA + YEH
	201	\u0650 → e ; # ARABIC KASRA
	202	\u064Fو → ū ; # ARABIC DAMMA + WAW
	203	\u064F → o ; # ARABIC DAMMA
51004dcb	204	\u0652 → ; # ARABIC SUKUN
729e4ab9	205	::NFC (NFD) ;
2ca993e8 A	206	#
	207	#
	208	########################################################################
	209