common/transforms/und_FONIPA-ar.xml

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE supplementalData SYSTEM "../../common/dtd/ldmlSupplemental.dtd">
<!--
Copyright © 1991-2016 Unicode, Inc.
CLDR data files are interpreted according to the LDML specification (http://unicode.org/reports/tr35/)
For terms of use, see http://www.unicode.org/copyright.html
-->
<supplementalData>
	<version number="$Revision$"/>
	<transforms>
		<transform source="und_FONIPA" target="ar" direction="forward" alias="ar-t-und-fonipa">
		  <tRule><![CDATA[
# 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 {e̞}];
$UVowel = [y {ɨ} {ʉ} ɯ u ʏ {ɪ̈} {ʊ̈} {ɯ̽} {ʊ} ø ɤ o {ø̞} {ɤ̞} {o̞} ɞ ɔ w {w̥} ʍ ʷ];
$AVowel = [ɛ œ ɜ ʌ æ ɐ a ɶ {ä} {ɒ̈} ɑ ɒ];
$SchwaVowel = [ɘ ɵ ə {ɵ̞}];
$Vowel = [$IVowel $UVowel $AVowel $SchwaVowel];
$Click = [ʘ ɋ ǀ ʇ ǃ ʗ ǂ ʄ ǁ ʖ];
$Boundary =  [^[:L:][:M:][:N:]];

::NFD;
[ʰ ʱ ʼ  ̃  ̰  ̋  ́  ̄  ̀  ̏  ̌  ̂ ˥ ˦ ˧ ˨ ˩ ꜜ ꜛ ↗ ↘  ͡  ͜  ̯] → ;
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 {e̞} $SchwaVowel] → uia;

# Kazakh Аягөз [ɑjɑɡy̯ʉz] → [ɑjɑɡiuz], to emit TODO
yʉ → iu;

::NULL;

# Vowels
$Boundary {ʔ? $IVowel ː} → إِي;
$Boundary {ʔ? $IVowel} → إِ;
{$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} → إِي;
$Boundary {ʔ? $SchwaVowel} → أ;
$SchwaVowel ː → ي;
$SchwaVowel → ;

# TODO: Handle glottal stop.
ʔ → ;

# Shadda for long (geminated) consonants
ː → ّ;

# Affricates
[{t͡ʃ} ʧ] → تْش;

# Clicks
[ɡ g ɠ k] $Click → كْش;
$Click → تْش;

# Nasal stops
[{m̥} m ɱ] → م;
[{n̼̊} {n̼} {n̥} n {ɳ̊} ɳ {ɲ̊} {ɲ̥} ɲ] → ن;
[{ŋ̊} ŋ {ɴ̥} ɴ] k → نك;
[{ŋ̊} ŋ {ɴ̥} ɴ] [ɡ g ɠ]? → نْغ;

# Non-nasal stops
[p b {p̪} {b̪} ɓ] → ب;
[{d̼} d ɗ ᶑ] → د;
[{t̼} t] → ت;
[ʈ] → ط;
[ɖ] → ض;
c → تْش;
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;
[{θ̼} θ {θ̱}] → ث;
[{ð̼} ð {ð̠}] → ذ;
ç → ش;
ʝ $IVowel? ː? → ي;
[x χ] → خ;
[ɣ ʁ] → غ;
ħ → ح;
ʕ → ع;
[h ɦ {ʔ̞}] → ه;

# Approximants, trills, flaps
ʋ → و;
ʙ → بر;
{r̝} → رش;
[{ɹ̥} {ɹ} {ɻ̊} {ɻ} {ɾ̥} ɾ {ɽ̊} ɽ {r̼} {r̥} r] → ر;
[{ʀ̥} ʀ] → غ;
ʜ → ح;
ʢ → ع;
j $IVowel? ː? → ي;

# Laterals
ɬ → شْل;
ɮ → جْل;
{[{ʎ̥} ʎ]} [^ $IVowel j ʝ] → لي;
[{l̼} {l̥} l {ɭ̊} ɭ {ʎ̥} ʎ] → ل;
[ʟ {ʟ̠}] → 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;
			]]></tRule>
		</transform>
	</transforms>
</supplementalData>