icuSources/data/translit/Grek_Latn_UNGEGN.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: Grek_Latn_UNGEGN.txt
   5 # Generated from CLDR
   6 #
   7
   8 # For modern Greek, based on UNGEGN rules.
   9 # Rules are predicated on running NFD first, and NFC afterwards
  10 # MINIMAL FILTER GENERATED FOR: Greek-Latin/UNGEGN
  11 # WARNING: need to add accents to both filters ###
  12 # :: [\u0301\u0304\u0306\u0308;µ·ÀÂÈÊÌÎÒÔÙÛàâèêìîòôùûĈ-ĉĜ-ĝĤ-ĥĴ-ĵŜ-ŝŴ-ŷǛ-ǜǸ-ǹ\u0300\u0302\u0313-\u0314\u0340\u0342-\u0343\u0345ͺ;Ά-ΊΌΎ-ΡΣ-ώϐ-ϖϰ-ϵЀЍѐѝḔ-ḕṐ-ṑẀ-ẁẐ-ẑẤ-ậẰ-ằẾ-ệỐ-ộỜ-ờỪ-ừỲ-ỳἀ-ἕἘ-Ἕἠ-ὅὈ-Ὅὐ-ὗὙὛὝὟ-ώᾀ-ᾴᾶ-ᾼι῁-ῄῆ-῍῏-ΐῖ-Ί῝῟-῭ῲ-ῴῶ-ῼΩϷ-\u07FBϹ] ;
  13 :: [[[:Greek:][:Mn:][:Me:]] [\:-;?·;·]] ;
  14 ::NFD (NFC) ;
  15 # Useful variables
  16 $lower = [[:latin:][:greek:] & [:Ll:]] ;
  17 $upper = [[:latin:][:greek:] & [:Lu:]] ;
  18 $accent = [[:Mn:][:Me:]] ;
  19 $macron = \u0304 ;
  20 $ddot = \u0308 ;
  21 $lcgvowel = [αεηιουω] ;
  22 $ucgvowel = [ΑΕΗΙΟΥΩ] ;
  23 $gvowel = [$lcgvowel $ucgvowel] ;
  24 $lcgvowelC = [$lcgvowel $accent] ;
  25 $evowel = [aeiouyAEIOUY];
  26 $vowel = [ $evowel $gvowel] ;
  27 $beforeLower = $accent * $lower ;
  28 $gammaLike = [ΓΚΞΧγκξχϰ] ;
  29 $egammaLike = [GKXCgkxc] ;
  30 $smooth = \u0313 ;
  31 $rough = \u0314 ;
  32 $iotasub = \u0345 ;
  33 $softener = [βΒγΓδΔζΖλΛμΜνΝρΡ$gvowel] ;
  34 $under = \u0331;
  35 $caron = \u030C;
  36 $afterLetter = [:L:] [\'$accent]* ;
  37 $beforeLetter = [\'$accent]* [:L:] ;
  38 # Fix punctuation
  39 # preserve orginal
  40 \: ↔ \: $under ;
  41 \? ↔ \? $under ;
  42 \; ↔ \? ;
  43 · ↔ \: ;
  44 # Fix any ancient characters that creep in
  45 \u0342 → \u0301 ;
  46 \u0302 → \u0301 ;
  47 \u0300 → \u0301 ;
  48 $smooth → ;
  49 $rough → ;
  50 $iotasub → ;
  51 ͺ → ;
  52 # need to have these up here so the rules don't mask
  53 η ↔ i $under ;
  54 Η ↔ I $under ;
  55 Ψ } $beforeLower ↔ Ps ;
  56 Ψ ↔ PS ;
  57 ψ ↔ ps ;
  58 ω ↔ o $under ;
  59 Ω ↔  O $under;
  60 # at begining or end of word, convert mp to b
  61 [^[:L:]$accent] { μπ → b ;
  62 μπ } [^[:L:]$accent] → b ;
  63 [^[:L:]$accent] { [Μμ][Ππ] → B ;
  64 [Μμ][Ππ] } [^[:L:]$accent] → B ;
  65 μπ ← b ;
  66 Μπ ← B } $beforeLower ;
  67 ΜΠ ← B ;
  68 # handle diphthongs ending with upsilon
  69 ου ↔ ou ;
  70 ΟΥ ↔ OU ;
  71 Ου ↔ Ou ;
  72 οΥ ↔ oU ;
  73 $fmaker = [aeiAEI] $under ? ;
  74 $shiftForwardVowels = [[:Mn:]-[\u0308]]; # note: a diaeresis keeps the items separate
  75 $fmaker { υ ( $shiftForwardVowels )* } $softener → $1 v $under ;
  76 υ $1 ← ( $shiftForwardVowels )* v $under ;
  77 $fmaker { υ ( $shiftForwardVowels )* } → $1 f $under;
  78 υ $1 ← ( $shiftForwardVowels )* f $under ;
  79 $fmaker { Υ } $softener ↔ V $under ;
  80 $fmaker { Υ ↔ U $under ;
  81 υ ↔ y ;
  82 Υ ↔ Y ;
  83 # NORMAL
  84 α ↔ a ;
  85 Α ↔ A ;
  86 β ↔ v ;
  87 Β ↔ V ;
  88 γ } $gammaLike ↔ n } $egammaLike ;
  89 γ ↔ g ;
  90 Γ } $gammaLike ↔ N } $egammaLike ;
  91 Γ ↔ G ;
  92 δ ↔ d ;
  93 Δ ↔ D ;
  94 ε ↔ e ;
  95 Ε ↔ E ;
  96 ζ ↔ z ;
  97 Ζ ↔ Z ;
  98 θ ↔ th ;
  99 Θ } $beforeLower ↔ Th ;
 100 Θ ↔ TH ;
 101 ι ↔ i ;
 102 Ι ↔ I ;
 103 κ ↔ k ;
 104 Κ ↔ K ;
 105 λ ↔ l ;
 106 Λ ↔ L ;
 107 μ ↔ m ;
 108 Μ ↔ M ;
 109 ν } $gammaLike → n\' ;
 110 ν ↔ n ;
 111 Ν } $gammaLike ↔ N\' ;
 112 Ν ↔ N ;
 113 ξ ↔ x ;
 114 Ξ ↔ X ;
 115 ο ↔ o ;
 116 Ο ↔ O ;
 117 π ↔ p ;
 118 Π ↔ P ;
 119 ρ ↔ r ;
 120 Ρ ↔ R ;
 121 # insert separator before things that turn into s
 122 [Pp] { } [ςσΣϷϸϺϻ] → \' ;
 123 # special S variants
 124 Ϸ ↔ S\u030C ; # Ϸ GREEK CAPITAL LETTER SHO Uppercase_Letter Grek - L
 125 ϸ ↔ s\u030C ; #ϸ GREEK SMALL LETTER SHO Lowercase_Letter Grek - L
 126 Ϻ ↔ S\u0302 ; # Ϻ GREEK CAPITAL LETTER SAN Uppercase_Letter Grek - L
 127 ϻ ↔ s\u0302 ; # ϻ GREEK SMALL LETTER SAN Lowercase_Letter Grek - L
 128 # Caron means exception
 129 # before a letter, initial
 130 ς } $beforeLetter ↔ s $under } $beforeLetter;
 131 σ } $beforeLetter ↔ s } $beforeLetter;
 132 # otherwise, after a letter = final
 133 $afterLetter { σ ↔ $afterLetter { s $under;
 134 $afterLetter { ς ↔ $afterLetter { s ;
 135 # otherwise (isolated) = initial
 136 ς ↔ s $under;
 137 σ ↔ s ;
 138 # [Pp] { Σ ↔ \'S ;
 139 Σ ↔ S ;
 140 τ ↔ t ;
 141 Τ ↔ T ;
 142 φ ↔ f ;
 143 Φ ↔ F ;
 144 χ ↔ ch ;
 145 Χ } $beforeLower ↔ Ch ;
 146 Χ ↔ CH ;
 147 # Completeness for ASCII
 148 # $ignore = [[:Mark:]''] * ;
 149 | ch ← h ;
 150 | k  ← c ;
 151 | i  ← j ;
 152 | k ← q ;
 153 | b ← u } $vowel ;
 154 | b ← w } $vowel ;
 155 | y ← u ;
 156 | y ← w ;
 157 | Ch ← H ;
 158 | K ← C ;
 159 | I ← J ;
 160 | K ← Q ;
 161 | B ← W } $vowel ;
 162 | B ← U } $vowel ;
 163 | Y ← W ;
 164 | Y ← U ;
 165 # Completeness for Greek
 166 ϐ → | β ;
 167 ϑ → | θ ;
 168 ϒ → | Υ ;
 169 ϕ → | φ ;
 170 ϖ → | π ;
 171 ϰ → | κ ;
 172 ϱ → | ρ ;
 173 ϲ → | σ ;
 174 Ϲ → | Σ; #U+03F9 GREEK CAPITAL LUNATE SIGMA SYMBOL
 175 ϳ → j ;
 176 ϴ → | Θ ;
 177 ϵ → | ε ;
 178 µ → | μ ;
 179 # delete any trailing ' marks used for roundtripping
 180 ← [Ππ] { \' } [Ss] ;
 181 ← [Νν] { \' } $egammaLike ;
 182 ::NFC (NFD) ;
 183 # MINIMAL FILTER GENERATED FOR: Latin-Greek/UNGEGN BACKWARD
 184 :: ([[[:Latin:][:Mn:][:Me:]] ['\:?]]) ;
 185