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

# ***************************************************************************
# *
# *  Copyright (C) 2004-2016, International Business Machines
# *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
# *
# ***************************************************************************
# File: und_FONIPA_ar.txt
# Generated from CLDR 
#

# Vowels
# ------
# In these rules, we produce ي و ا both for short and for long vowels.
# This would be wrong for writing Arabic, but when transliterating
# foreign words and names, it is strongly preferred to vowel marks.
# However, we emit short schwa [ə] and a few other, schwa-like vowels.
$IVowel = [i ɪ e {e\u031E}];
$UVowel = [y {ɨ} {ʉ} ɯ u ʏ {ɪ\u0308} {ʊ\u0308} {ɯ\u033D} {ʊ} ø ɤ o {ø\u031E} {ɤ\u031E} {o\u031E} ɞ ɔ w {w\u0325} ʍ ʷ];
$AVowel = [ɛ œ ɜ ʌ æ ɐ a ɶ {ä} {ɒ\u0308} ɑ ɒ];
$SchwaVowel = [ɘ ɵ ə {ɵ\u031E}];
$Vowel = [$IVowel $UVowel $AVowel $SchwaVowel];
$Click = [ʘ ɋ ǀ ʇ ǃ ʗ ǂ ʄ ǁ ʖ];
$Boundary =  [^[:L:][:M:][:N:]];
::NFD;
[ʰ ʱ ʼ  \u0303  \u0330  \u030B  \u0301  \u0304  \u0300  \u030F  \u030C  \u0302 ˥ ˦ ˧ ˨ ˩ ꜜ ꜛ ↗ ↘  \u0361  \u035C  \u032F] → ;
ʲ → j;
ᵐ → m;
ⁿ → n;
ᵑ → ŋ;
::NFC;
# TODO: Diphthongs probably need more work.
# Romanian [sekujesk] → [sekujask], for emitting سيكوياسك not سيكويسك
$UVowel [j $IVowel] [e {e\u031E} $SchwaVowel] → uia;
# Kazakh Аягөз [ɑjɑɡy\u032Fʉz] → [ɑjɑɡiuz], to emit TODO
yʉ → iu;
::NULL;
# Vowels
$Boundary {ʔ? $IVowel ː} → إ\u0650ي;
$Boundary {ʔ? $IVowel} → إ\u0650;
{$IVowel ʔ} $Boundary → ئ;
{$IVowel ː ʔ} $Boundary → يء;
{$IVowel ː ʔ} [$Vowel] → ئ;
$IVowel ː? → ي;
$Boundary {ʔ? $UVowel ː} → أو;
$Boundary {ʔ? $UVowel} → أ;
{$UVowel ʔ} $Boundary → ؤ;
{$UVowel ː ʔ} $Boundary → وء;
$UVowel ː? → و;
$Boundary {ʔ? $AVowel ː} → آ;
$Boundary {ʔ? $AVowel} → أ;
{$AVowel ʔ} $Boundary → أ;
{$AVowel ː ʔ} $Boundary → اء;
$AVowel ː? ʔ $AVowel ː? → اءا;
$AVowel ː? → ا;
$Boundary {ʔ? $SchwaVowel ː} → إ\u0650ي;
$Boundary {ʔ? $SchwaVowel} → أ;
$SchwaVowel ː → ي;
$SchwaVowel → ;
# TODO: Handle glottal stop.
ʔ → ;
# Shadda for long (geminated) consonants
ː → \u0651;
# Affricates
[{t\u0361ʃ} ʧ] → ت\u0652ش;
# Clicks
[ɡ g ɠ k] $Click → ك\u0652ش;
$Click → ت\u0652ش;
# Nasal stops
[{m\u0325} m ɱ] → م;
[{n\u033C\u030A} {n\u033C} {n\u0325} n {ɳ\u030A} ɳ {ɲ\u030A} {ɲ\u0325} ɲ] → ن;
[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] k → نك;
[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] [ɡ g ɠ]? → ن\u0652غ;
# Non-nasal stops
[p b {p\u032A} {b\u032A} ɓ] → ب;
[{d\u033C} d ɗ ᶑ] → د;
[{t\u033C} t] → ت;
[ʈ] → ط;
[ɖ] → ض;
c → ت\u0652ش;
ɟ → دج;
k → ك;
[ɡ g ɠ] → غ;
[q ɢ ʡ ʛ] → ق;
# Sibilant fricatives
s → س;
z → ز;
[ʃ ʂ ɕ ʄ] → ش;
[ʒ ʐ ʑ] → ج;
# Non-sibilant fricatives
[ɸ f v] → ف;
β → ب;
[{θ\u033C} θ {θ\u0331}] → ث;
[{ð\u033C} ð {ð\u0320}] → ذ;
ç → ش;
ʝ $IVowel? ː? → ي;
[x χ] → خ;
[ɣ ʁ] → غ;
ħ → ح;
ʕ → ع;
[h ɦ {ʔ\u031E}] → ه;
# Approximants, trills, flaps
ʋ → و;
ʙ → بر;
{r\u031D} → رش;
[{ɹ\u0325} {ɹ} {ɻ\u030A} {ɻ} {ɾ\u0325} ɾ {ɽ\u030A} ɽ {r\u033C} {r\u0325} r] → ر;
[{ʀ\u0325} ʀ] → غ;
ʜ → ح;
ʢ → ع;
j $IVowel? ː? → ي;
# Laterals
ɬ → ش\u0652ل;
ɮ → ج\u0652ل;
{[{ʎ\u0325} ʎ]} [^ $IVowel j ʝ] → لي;
[{l\u033C} {l\u0325} l {ɭ\u030A} ɭ {ʎ\u0325} ʎ] → ل;
[ʟ {ʟ\u0320}] → غ;
# Independent pass for misc cleanup.
::NULL;
# Strip off syllable markers
\. → ;
# Sequences of three or more ووو look very confusing; we shorten them.
# Polish Darłowo [darwɔvɔ] → داروو → داروووو
ووو+ → وو;
Commit	Line	Data
2ca993e8 A	1	# ***************************************************************************
	2	# *
	3	# * Copyright (C) 2004-2016, International Business Machines
	4	# * Corporation; Unicode, Inc.; and others. All Rights Reserved.
	5	# *
	6	# ***************************************************************************
	7	# File: und_FONIPA_ar.txt
	8	# Generated from CLDR
	9	#
	10
	11	# Vowels
	12	# ------
	13	# In these rules, we produce ي و ا both for short and for long vowels.
	14	# This would be wrong for writing Arabic, but when transliterating
	15	# foreign words and names, it is strongly preferred to vowel marks.
	16	# However, we emit short schwa [ə] and a few other, schwa-like vowels.
	17	$IVowel = [i ɪ e {e\u031E}];
	18	$UVowel = [y {ɨ} {ʉ} ɯ u ʏ {ɪ\u0308} {ʊ\u0308} {ɯ\u033D} {ʊ} ø ɤ o {ø\u031E} {ɤ\u031E} {o\u031E} ɞ ɔ w {w\u0325} ʍ ʷ];
	19	$AVowel = [ɛ œ ɜ ʌ æ ɐ a ɶ {ä} {ɒ\u0308} ɑ ɒ];
	20	$SchwaVowel = [ɘ ɵ ə {ɵ\u031E}];
	21	$Vowel = [$IVowel $UVowel $AVowel $SchwaVowel];
	22	$Click = [ʘ ɋ ǀ ʇ ǃ ʗ ǂ ʄ ǁ ʖ];
	23	$Boundary = [^[:L:][:M:][:N:]];
	24	::NFD;
	25	[ʰ ʱ ʼ \u0303 \u0330 \u030B \u0301 \u0304 \u0300 \u030F \u030C \u0302 ˥ ˦ ˧ ˨ ˩ ꜜ ꜛ ↗ ↘ \u0361 \u035C \u032F] → ;
	26	ʲ → j;
	27	ᵐ → m;
	28	ⁿ → n;
	29	ᵑ → ŋ;
	30	::NFC;
	31	# TODO: Diphthongs probably need more work.
	32	# Romanian [sekujesk] → [sekujask], for emitting سيكوياسك not سيكويسك
	33	$UVowel [j $IVowel] [e {e\u031E} $SchwaVowel] → uia;
	34	# Kazakh Аягөз [ɑjɑɡy\u032Fʉz] → [ɑjɑɡiuz], to emit TODO
	35	yʉ → iu;
	36	::NULL;
	37	# Vowels
	38	$Boundary {ʔ? $IVowel ː} → إ\u0650ي;
	39	$Boundary {ʔ? $IVowel} → إ\u0650;
	40	{$IVowel ʔ} $Boundary → ئ;
	41	{$IVowel ː ʔ} $Boundary → يء;
	42	{$IVowel ː ʔ} [$Vowel] → ئ;
	43	$IVowel ː? → ي;
	44	$Boundary {ʔ? $UVowel ː} → أو;
	45	$Boundary {ʔ? $UVowel} → أ;
	46	{$UVowel ʔ} $Boundary → ؤ;
	47	{$UVowel ː ʔ} $Boundary → وء;
	48	$UVowel ː? → و;
	49	$Boundary {ʔ? $AVowel ː} → آ;
	50	$Boundary {ʔ? $AVowel} → أ;
	51	{$AVowel ʔ} $Boundary → أ;
	52	{$AVowel ː ʔ} $Boundary → اء;
	53	$AVowel ː? ʔ $AVowel ː? → اءا;
	54	$AVowel ː? → ا;
	55	$Boundary {ʔ? $SchwaVowel ː} → إ\u0650ي;
	56	$Boundary {ʔ? $SchwaVowel} → أ;
	57	$SchwaVowel ː → ي;
	58	$SchwaVowel → ;
	59	# TODO: Handle glottal stop.
	60	ʔ → ;
	61	# Shadda for long (geminated) consonants
	62	ː → \u0651;
	63	# Affricates
	64	[{t\u0361ʃ} ʧ] → ت\u0652ش;
65	# Clicks
66	[ɡ g ɠ k] $Click → ك\u0652ش;
67	$Click → ت\u0652ش;
68	# Nasal stops
69	[{m\u0325} m ɱ] → م;
70	[{n\u033C\u030A} {n\u033C} {n\u0325} n {ɳ\u030A} ɳ {ɲ\u030A} {ɲ\u0325} ɲ] → ن;
71	[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] k → نك;
72	[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] [ɡ g ɠ]? → ن\u0652غ;
73	# Non-nasal stops
74	[p b {p\u032A} {b\u032A} ɓ] → ب;
75	[{d\u033C} d ɗ ᶑ] → د;
76	[{t\u033C} t] → ت;
77	[ʈ] → ط;
78	[ɖ] → ض;
79	c → ت\u0652ش;
80	ɟ → دج;
81	k → ك;
82	[ɡ g ɠ] → غ;
83	[q ɢ ʡ ʛ] → ق;
84	# Sibilant fricatives
85	s → س;
86	z → ز;
87	[ʃ ʂ ɕ ʄ] → ش;
88	[ʒ ʐ ʑ] → ج;
89	# Non-sibilant fricatives
90	[ɸ f v] → ف;
91	β → ب;
92	[{θ\u033C} θ {θ\u0331}] → ث;
93	[{ð\u033C} ð {ð\u0320}] → ذ;
94	ç → ش;
95	ʝ $IVowel? ː? → ي;
96	[x χ] → خ;
97	[ɣ ʁ] → غ;
98	ħ → ح;
99	ʕ → ع;
100	[h ɦ {ʔ\u031E}] → ه;
101	# Approximants, trills, flaps
102	ʋ → و;
103	ʙ → بر;
104	{r\u031D} → رش;
105	[{ɹ\u0325} {ɹ} {ɻ\u030A} {ɻ} {ɾ\u0325} ɾ {ɽ\u030A} ɽ {r\u033C} {r\u0325} r] → ر;
106	[{ʀ\u0325} ʀ] → غ;
107	ʜ → ح;
108	ʢ → ع;
109	j $IVowel? ː? → ي;
110	# Laterals
111	ɬ → ش\u0652ل;
112	ɮ → ج\u0652ل;
113	{[{ʎ\u0325} ʎ]} [^ $IVowel j ʝ] → لي;
114	[{l\u033C} {l\u0325} l {ɭ\u030A} ɭ {ʎ\u0325} ʎ] → ل;
115	[ʟ {ʟ\u0320}] → غ;
116	# Independent pass for misc cleanup.
117	::NULL;
118	# Strip off syllable markers
119	\. → ;
120	# Sequences of three or more ووو look very confusing; we shorten them.
121	# Polish Darłowo [darwɔvɔ] → داروو → داروووو
122	ووو+ → وو;
123