icu4c/source/data/translit/und_FONIPA_ar.txt

# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html
# Generated using tools/cldr/cldr-to-icu/build-icu-data.xml
#
# 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;