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

# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html#License
#
# 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
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: und_FONIPA_ar.txt
f3c0d7a5	5	# Generated from CLDR
2ca993e8 A	6	#
	7
	8	# Vowels
	9	# ------
	10	# In these rules, we produce ي و ا both for short and for long vowels.
	11	# This would be wrong for writing Arabic, but when transliterating
	12	# foreign words and names, it is strongly preferred to vowel marks.
	13	# However, we emit short schwa [ə] and a few other, schwa-like vowels.
	14	$IVowel = [i ɪ e {e\u031E}];
	15	$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];
	16	$AVowel = [ɛ œ ɜ ʌ æ ɐ a ɶ {ä} {ɒ\u0308} ɑ ɒ];
	17	$SchwaVowel = [ɘ ɵ ə {ɵ\u031E}];
	18	$Vowel = [$IVowel $UVowel $AVowel $SchwaVowel];
	19	$Click = [ʘ ɋ ǀ ʇ ǃ ʗ ǂ ʄ ǁ ʖ];
	20	$Boundary = [^[:L:][:M:][:N:]];
	21	::NFD;
	22	[ʰ ʱ ʼ \u0303 \u0330 \u030B \u0301 \u0304 \u0300 \u030F \u030C \u0302 ˥ ˦ ˧ ˨ ˩ ꜜ ꜛ ↗ ↘ \u0361 \u035C \u032F] → ;
	23	ʲ → j;
	24	ᵐ → m;
	25	ⁿ → n;
	26	ᵑ → ŋ;
	27	::NFC;
	28	# TODO: Diphthongs probably need more work.
	29	# Romanian [sekujesk] → [sekujask], for emitting سيكوياسك not سيكويسك
	30	$UVowel [j $IVowel] [e {e\u031E} $SchwaVowel] → uia;
	31	# Kazakh Аягөз [ɑjɑɡy\u032Fʉz] → [ɑjɑɡiuz], to emit TODO
	32	yʉ → iu;
	33	::NULL;
	34	# Vowels
	35	$Boundary {ʔ? $IVowel ː} → إ\u0650ي;
	36	$Boundary {ʔ? $IVowel} → إ\u0650;
	37	{$IVowel ʔ} $Boundary → ئ;
	38	{$IVowel ː ʔ} $Boundary → يء;
	39	{$IVowel ː ʔ} [$Vowel] → ئ;
	40	$IVowel ː? → ي;
	41	$Boundary {ʔ? $UVowel ː} → أو;
	42	$Boundary {ʔ? $UVowel} → أ;
	43	{$UVowel ʔ} $Boundary → ؤ;
	44	{$UVowel ː ʔ} $Boundary → وء;
	45	$UVowel ː? → و;
	46	$Boundary {ʔ? $AVowel ː} → آ;
	47	$Boundary {ʔ? $AVowel} → أ;
	48	{$AVowel ʔ} $Boundary → أ;
	49	{$AVowel ː ʔ} $Boundary → اء;
	50	$AVowel ː? ʔ $AVowel ː? → اءا;
	51	$AVowel ː? → ا;
	52	$Boundary {ʔ? $SchwaVowel ː} → إ\u0650ي;
	53	$Boundary {ʔ? $SchwaVowel} → أ;
	54	$SchwaVowel ː → ي;
	55	$SchwaVowel → ;
	56	# TODO: Handle glottal stop.
	57	ʔ → ;
	58	# Shadda for long (geminated) consonants
	59	ː → \u0651;
	60	# Affricates
	61	[{t\u0361ʃ} ʧ] → ت\u0652ش;
	62	# Clicks
	63	[ɡ g ɠ k] $Click → ك\u0652ش;
	64	$Click → ت\u0652ش;
	65	# Nasal stops
	66	[{m\u0325} m ɱ] → م;
	67	[{n\u033C\u030A} {n\u033C} {n\u0325} n {ɳ\u030A} ɳ {ɲ\u030A} {ɲ\u0325} ɲ] → ن;
	68	[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] k → نك;
	69	[{ŋ\u030A} ŋ {ɴ\u0325} ɴ] [ɡ g ɠ]? → ن\u0652غ;
70	# Non-nasal stops
71	[p b {p\u032A} {b\u032A} ɓ] → ب;
72	[{d\u033C} d ɗ ᶑ] → د;
73	[{t\u033C} t] → ت;
74	[ʈ] → ط;
75	[ɖ] → ض;
76	c → ت\u0652ش;
77	ɟ → دج;
78	k → ك;
79	[ɡ g ɠ] → غ;
80	[q ɢ ʡ ʛ] → ق;
81	# Sibilant fricatives
82	s → س;
83	z → ز;
84	[ʃ ʂ ɕ ʄ] → ش;
85	[ʒ ʐ ʑ] → ج;
86	# Non-sibilant fricatives
87	[ɸ f v] → ف;
88	β → ب;
89	[{θ\u033C} θ {θ\u0331}] → ث;
90	[{ð\u033C} ð {ð\u0320}] → ذ;
91	ç → ش;
92	ʝ $IVowel? ː? → ي;
93	[x χ] → خ;
94	[ɣ ʁ] → غ;
95	ħ → ح;
96	ʕ → ع;
97	[h ɦ {ʔ\u031E}] → ه;
98	# Approximants, trills, flaps
99	ʋ → و;
100	ʙ → بر;
101	{r\u031D} → رش;
102	[{ɹ\u0325} {ɹ} {ɻ\u030A} {ɻ} {ɾ\u0325} ɾ {ɽ\u030A} ɽ {r\u033C} {r\u0325} r] → ر;
103	[{ʀ\u0325} ʀ] → غ;
104	ʜ → ح;
105	ʢ → ع;
106	j $IVowel? ː? → ي;
107	# Laterals
108	ɬ → ش\u0652ل;
109	ɮ → ج\u0652ل;
110	{[{ʎ\u0325} ʎ]} [^ $IVowel j ʝ] → لي;
111	[{l\u033C} {l\u0325} l {ɭ\u030A} ɭ {ʎ\u0325} ʎ] → ل;
112	[ʟ {ʟ\u0320}] → غ;
113	# Independent pass for misc cleanup.
114	::NULL;
115	# Strip off syllable markers
116	\. → ;
117	# Sequences of three or more ووو look very confusing; we shorten them.
118	# Polish Darłowo [darwɔvɔ] → داروو → داروووو
119	ووو+ → وو;
120