icuSources/data/translit/und_FONIPA_fa.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: und_FONIPA_fa.txt
   5 # Generated from CLDR
   6 #
   7
   8 # Vowels
   9 # ------
  10 # In these rules, we produce ی و ا both for short and for long vowels.
  11 # This would be wrong for writing Farsi or Arabic, but when transliterating
  12 # foreign words and names, it is strongly preferred to vowel marks.
  13 # Short schwa [ə] and a few other, schwa-like vowels get omitted entirely
  14 # unless at the end of the word, in which case we emit ه whose Farsi
  15 # word-final pronunciation comes close to [ə]. At the beginning of words,
  16 # Farsi speakers prefer to see آ for [ɑ] and a few other similar-sounding
  17 # dark vowels; note that this use of آ is quite different from Arabic.
  18 $IVowel = [i ɪ e {e\u031E}];
  19 $UVowel = [y {ɨ} {ʉ} ɯ u ʏ {ɪ\u0308} {ʊ\u0308} {ɯ\u033D} {ʊ} ø ɤ o {ø\u031E} {ɤ\u031E} {o\u031E} ɔ w {w\u0325} ʍ ʷ];
  20 $AVowel = [ɛ œ ɜ æ ɶ];
  21 $DarkAVowel = [ʌ a ɑ ɒ ɐ ɞ {ä} {ɒ\u0308}];  # آ instead of ا at beginning of words
  22 $SchwaVowel = [ɘ ɵ ə {ɵ\u031E}];
  23 $Click = [ʘ ɋ ǀ ʇ ǃ ʗ ǂ ʄ ǁ ʖ];
  24 $Boundary =  [^[:L:][:M:][:N:]];
  25 ::NFD;
  26 [ʰ ʱ ʼ  \u0303  \u0330  \u030B  \u0301  \u0304  \u0300  \u030F  \u030C  \u0302 ˥ ˦ ˧ ˨ ˩ ꜜ ꜛ ↗ ↘  \u0361  \u035C  \u032F] → ;
  27 ʲ → j;
  28 ᵐ → m;
  29 ⁿ → n;
  30 ᵑ → ŋ;
  31 ::NFC;
  32 # TODO: Diphthongs probably need more work.
  33 # Romanian [sekujesk] → [sekujask], for emitting سیکویاسک not سیکویسک
  34 $UVowel [j $IVowel] [e {e\u031E} $SchwaVowel] → uia;
  35 # Kazakh Аягөз [ɑjɑɡy\u032Fʉz] → [ɑjɑɡiuz], to emit آیاگیوز not آیاگووز
  36 yʉ → iu;
  37 ::NULL;
  38 # Vowels
  39 $Boundary {$SchwaVowel ː?} → ای;
  40 $SchwaVowel ː → ی;
  41 {[$SchwaVowel e {e\u031E}]} [^[:L:][:M:][:N:][\.]] → ه;
  42 $SchwaVowel → ;
  43 $Boundary {$IVowel ː?} → ای;
  44 $IVowel ː? j? → ی;
  45 $Boundary {$UVowel ː?} → او;
  46 $UVowel ː? → و;
  47 $Boundary {$AVowel ː?} → ا;
  48 $AVowel ː? → ا;
  49 $Boundary {$DarkAVowel ː?} → آ;
  50 $DarkAVowel ː? → ا;
  51 # Shadda for long (geminated) consonants
  52 ː → \u0651;
  53 # Affricates
  54 [{t\u0361ʃ} ʧ] → چ;
  55 # Clicks
  56 [ɡ g ɠ k] $Click → کچ;
  57 [n ɲ]? $Click → نچ;
  58 # Nasal stops
  59 [{m\u0325} m ɱ] → م;
  60 [{n\u033C\u030A} {n\u033C} {n\u0325} n {ɳ\u030A} ɳ {ɲ\u030A} {ɲ\u0325} ɲ] → ن;
  61 [{ŋ\u030A} ŋ {ɴ\u0325} ɴ] k → نک;
  62 [{ŋ\u030A} ŋ {ɴ\u0325} ɴ] [ɡ g]? → نگ;
  63 # Non-nasal stops
  64 [p {p\u032A}] → پ;
  65 [b {b\u032A} ɓ] → ب;
  66 [{d\u033C} d ɗ ᶑ] → د;
  67 [{t\u033C} t] → ت;
  68 [ʈ] → ط;
  69 [ɖ] → ض;
  70 c → چ;
  71 ɟ → دج;
  72 k → ک;
  73 [ɡ g ɠ] → گ;
  74 [q ɢ ʡ ʛ] → ق;
  75 ʔ → ;
  76 # Sibilant fricatives
  77 s → س;
  78 z → ز;
  79 [ʃ ʂ ɕ ʄ] → ش;
  80 [ʒ ʐ ʑ] → ژ;
  81 # Non-sibilant fricatives
  82 [ɸ f] → ف;
  83 [β v] → و;
  84 [{θ\u033C} θ {θ\u0331}] → ث;
  85 [{ð\u033C} ð {ð\u0320}] → ذ;
  86 ç → ش;
  87 ʝ $IVowel? ː? → ی;
  88 [x χ] → خ;
  89 [ɣ ʁ] → غ;
  90 ħ → ح;
  91 ʕ → ع;
  92 [h ɦ {ʔ\u031E}] → ه;
  93 # Approximants, trills, flaps
  94 ʋ → و;
  95 ʙ → بر;
  96 {r\u031D} → رژ;
  97 [{ɹ\u0325} {ɹ} {ɻ\u030A} {ɻ} {ɾ\u0325} ɾ {ɽ\u030A} ɽ {r\u033C} {r\u0325} r] → ر;
  98 [{ʀ\u0325} ʀ] → غ;
  99 ʜ → ح;
 100 ʢ → ع;
 101 j $IVowel? ː? → ی;
 102 # Laterals
 103 ɬ → شل;
 104 ɮ → ژل;
 105 {[{ʎ\u0325} ʎ]} [^ $IVowel j ʝ] → لی;
 106 [{l\u033C} {l\u0325} l {ɭ\u030A} ɭ {ʎ\u0325} ʎ] → ل;
 107 [ʟ {ʟ\u0320}] → غ;
 108 # Independent pass for misc cleanup.
 109 ::NULL;
 110 # Strip off syllable markers
 111 \. → ;
 112 # Sequences of three or more ووو look very confusing; we shorten them.
 113 # Polish Darłowo [darwɔvɔ] → داروو → داروووو
 114 ووو+ → وو;
 115