icuSources/data/translit/Grek_Latn_UNGEGN.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Grek_Latn_UNGEGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 # For modern Greek, based on UNGEGN rules.
  12 # Rules are predicated on running NFD first, and NFC afterwards
  13 # MINIMAL FILTER GENERATED FOR: Greek-Latin/UNGEGN
  14 # WARNING: need to add accents to both filters ###
  15 # :: [\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] ;
  16 :: [[[:Greek:][:Mn:][:Me:]] [\:-;?·;·]] ;
  17 ::NFD (NFC) ;
  18 # Useful variables
  19 $lower = [[:latin:][:greek:] & [:Ll:]] ;
  20 $upper = [[:latin:][:greek:] & [:Lu:]] ;
  21 $accent = [[:Mn:][:Me:]] ;
  22 $macron = \u0304 ;
  23 $ddot = \u0308 ;
  24 $lcgvowel = [αεηιουω] ;
  25 $ucgvowel = [ΑΕΗΙΟΥΩ] ;
  26 $gvowel = [$lcgvowel $ucgvowel] ;
  27 $lcgvowelC = [$lcgvowel $accent] ;
  28 $evowel = [aeiouyAEIOUY];
  29 $vowel = [ $evowel $gvowel] ;
  30 $beforeLower = $accent * $lower ;
  31 $gammaLike = [ΓΚΞΧγκξχϰ] ;
  32 $egammaLike = [GKXCgkxc] ;
  33 $smooth = \u0313 ;
  34 $rough = \u0314 ;
  35 $iotasub = \u0345 ;
  36 $softener = [βΒγΓδΔζΖλΛμΜνΝρΡ$gvowel] ;
  37 $under = \u0331;
  38 $caron = \u030C;
  39 $afterLetter = [:L:] [\'$accent]* ;
  40 $beforeLetter = [\'$accent]* [:L:] ;
  41 # Fix punctuation
  42 # preserve orginal
  43 \: ↔ \: $under ;
  44 \? ↔ \? $under ;
  45 \; ↔ \? ;
  46 · ↔ \: ;
  47 # Fix any ancient characters that creep in
  48 \u0342 → \u0301 ;
  49 \u0302 → \u0301 ;
  50 \u0300 → \u0301 ;
  51 $smooth → ;
  52 $rough → ;
  53 $iotasub → ;
  54 ͺ → ;
  55 # need to have these up here so the rules don't mask
  56 η ↔ i $under ;
  57 Η ↔ I $under ;
  58 Ψ } $beforeLower ↔ Ps ;
  59 Ψ ↔ PS ;
  60 ψ ↔ ps ;
  61 ω ↔ o $under ;
  62 Ω ↔  O $under;
  63 # at begining or end of word, convert mp to b
  64 [^[:L:]$accent] { μπ → b ;
  65 μπ } [^[:L:]$accent] → b ;
  66 [^[:L:]$accent] { [Μμ][Ππ] → B ;
  67 [Μμ][Ππ] } [^[:L:]$accent] → B ;
  68 μπ ← b ;
  69 Μπ ← B } $beforeLower ;
  70 ΜΠ ← B ;
  71 # handle diphthongs ending with upsilon
  72 ου ↔ ou ;
  73 ΟΥ ↔ OU ;
  74 Ου ↔ Ou ;
  75 οΥ ↔ oU ;
  76 $fmaker = [aeiAEI] $under ? ;
  77 $shiftForwardVowels = [[:Mn:]-[\u0308]]; # note: a diaeresis keeps the items separate
  78 $fmaker { υ ( $shiftForwardVowels )* } $softener → $1 v $under ;
  79 υ $1 ← ( $shiftForwardVowels )* v $under ;
  80 $fmaker { υ ( $shiftForwardVowels )* } → $1 f $under;
  81 υ $1 ← ( $shiftForwardVowels )* f $under ;
  82 $fmaker { Υ } $softener ↔ V $under ;
  83 $fmaker { Υ ↔ U $under ;
  84 υ ↔ y ;
  85 Υ ↔ Y ;
  86 # NORMAL
  87 α ↔ a ;
  88 Α ↔ A ;
  89 β ↔ v ;
  90 Β ↔ V ;
  91 γ } $gammaLike ↔ n } $egammaLike ;
  92 γ ↔ g ;
  93 Γ } $gammaLike ↔ N } $egammaLike ;
  94 Γ ↔ G ;
  95 δ ↔ d ;
  96 Δ ↔ D ;
  97 ε ↔ e ;
  98 Ε ↔ E ;
  99 ζ ↔ z ;
 100 Ζ ↔ Z ;
 101 θ ↔ th ;
 102 Θ } $beforeLower ↔ Th ;
 103 Θ ↔ TH ;
 104 ι ↔ i ;
 105 Ι ↔ I ;
 106 κ ↔ k ;
 107 Κ ↔ K ;
 108 λ ↔ l ;
 109 Λ ↔ L ;
 110 μ ↔ m ;
 111 Μ ↔ M ;
 112 ν } $gammaLike → n\' ;
 113 ν ↔ n ;
 114 Ν } $gammaLike ↔ N\' ;
 115 Ν ↔ N ;
 116 ξ ↔ x ;
 117 Ξ ↔ X ;
 118 ο ↔ o ;
 119 Ο ↔ O ;
 120 π ↔ p ;
 121 Π ↔ P ;
 122 ρ ↔ r ;
 123 Ρ ↔ R ;
 124 # insert separator before things that turn into s
 125 [Pp] { } [ςσΣϷϸϺϻ] → \' ;
 126 # special S variants
 127 Ϸ ↔ S\u030C ; # Ϸ GREEK CAPITAL LETTER SHO Uppercase_Letter Grek - L
 128 ϸ ↔ s\u030C ; #ϸ GREEK SMALL LETTER SHO Lowercase_Letter Grek - L
 129 Ϻ ↔ S\u0302 ; # Ϻ GREEK CAPITAL LETTER SAN Uppercase_Letter Grek - L
 130 ϻ ↔ s\u0302 ; # ϻ GREEK SMALL LETTER SAN Lowercase_Letter Grek - L
 131 # Caron means exception
 132 # before a letter, initial
 133 ς } $beforeLetter ↔ s $under } $beforeLetter;
 134 σ } $beforeLetter ↔ s } $beforeLetter;
 135 # otherwise, after a letter = final
 136 $afterLetter { σ ↔ $afterLetter { s $under;
 137 $afterLetter { ς ↔ $afterLetter { s ;
 138 # otherwise (isolated) = initial
 139 ς ↔ s $under;
 140 σ ↔ s ;
 141 # [Pp] { Σ ↔ \'S ;
 142 Σ ↔ S ;
 143 τ ↔ t ;
 144 Τ ↔ T ;
 145 φ ↔ f ;
 146 Φ ↔ F ;
 147 χ ↔ ch ;
 148 Χ } $beforeLower ↔ Ch ;
 149 Χ ↔ CH ;
 150 # Completeness for ASCII
 151 # $ignore = [[:Mark:]''] * ;
 152 | ch ← h ;
 153 | k  ← c ;
 154 | i  ← j ;
 155 | k ← q ;
 156 | b ← u } $vowel ;
 157 | b ← w } $vowel ;
 158 | y ← u ;
 159 | y ← w ;
 160 | Ch ← H ;
 161 | K ← C ;
 162 | I ← J ;
 163 | K ← Q ;
 164 | B ← W } $vowel ;
 165 | B ← U } $vowel ;
 166 | Y ← W ;
 167 | Y ← U ;
 168 # Completeness for Greek
 169 ϐ → | β ;
 170 ϑ → | θ ;
 171 ϒ → | Υ ;
 172 ϕ → | φ ;
 173 ϖ → | π ;
 174 ϰ → | κ ;
 175 ϱ → | ρ ;
 176 ϲ → | σ ;
 177 Ϲ → | Σ; #U+03F9 GREEK CAPITAL LUNATE SIGMA SYMBOL
 178 ϳ → j ;
 179 ϴ → | Θ ;
 180 ϵ → | ε ;
 181 µ → | μ ;
 182 # delete any trailing ' marks used for roundtripping
 183 ← [Ππ] { \' } [Ss] ;
 184 ← [Νν] { \' } $egammaLike ;
 185 ::NFC (NFD) ;
 186 # MINIMAL FILTER GENERATED FOR: Latin-Greek/UNGEGN BACKWARD
 187 :: ([[[:Latin:][:Mn:][:Me:]] ['\:?]]) ;
 188