ICU-6.2.22.tar.gz

[apple/icu.git] / icuSources / data / translit / Greek_Latin_UNGEGN.txt

#--------------------------------------------------------------------\r
# Copyright (c) 1999-2004, International Business Machines\r
# Corporation and others. All Rights Reserved.\r
#--------------------------------------------------------------------\r
# For modern Greek, based on UNGEGN rules.\r
\r
# Rules are predicated on running NFD first, and NFC afterwards\r
# MINIMAL FILTER GENERATED FOR: Greek-Latin/UNGEGN\r
# WARNING: need to add accents to both filters ###\r
# :: [́̄̆̈;µ·ÀÂÈÊÌÎÒÔÙÛàâèêìîòôùûĈ-ĉĜ-ĝĤ-ĥĴ-ĵŜ-ŝŴ-ŷǛ-ǜǸ-ǹ̀̂̓-̔̀͂-̓ͅͺ;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-ῼΩ\u03F7-\u07FB\u03F9] ;\r
\r
:: [[[:Greek:][:Mn:][:Me:]] [\:-;?\u00B7\u037E\u0387]] ;\r
::NFD (NFC) ;\r
\r
# Useful variables\r
\r
$lower = [[:latin:][:greek:] & [:Ll:]] ;\r
$upper = [[:latin:][:greek:] & [:Lu:]] ;\r
$accent = [[:Mn:][:Me:]] ;\r
\r
$macron = ̄ ;\r
$ddot = ̈ ;\r
\r
$lcgvowel = [αεηιουω] ;\r
$ucgvowel = [ΑΕΗΙΟΥΩ] ;\r
$gvowel = [$lcgvowel $ucgvowel] ;\r
$lcgvowelC = [$lcgvowel $accent] ;\r
\r
$evowel = [aeiouyAEIOUY];\r
$vowel = [ $evowel $gvowel] ;\r
\r
$beforeLower = $accent * $lower ;\r
\r
$gammaLike = [ΓΚΞΧγκξχϰ] ;\r
$egammaLike = [GKXCgkxc] ;\r
$smooth = ̓ ;\r
$rough = ̔ ;\r
$iotasub = ͅ ;\r
\r
$softener = [βΒγΓδΔζΖλΛμΜνΝρΡ$gvowel] ;\r
\r
$under = ̱;\r
\r
$caron = ̌;\r
\r
$afterLetter = [:L:] [\'$accent]* ;\r
$beforeLetter = [\'$accent]* [:L:] ;\r
\r
# Fix punctuation\r
\r
# preserve orginal\r
\: <> \: $under ;\r
\? <> \? $under ;\r
\r
\; <> \? ;\r
· <> \: ;\r
\r
# Fix any ancient characters that creep in\r
\r
͂ > ́ ;\r
̂ > ́ ;\r
̀ > ́ ;\r
$smooth > ;\r
$rough > ;\r
$iotasub > ;\r
ͺ > ;\r
\r
# need to have these up here so the rules don't mask\r
\r
η <> i $under ;\r
Η <> I $under ;\r
\r
Ψ } $beforeLower <> Ps ;\r
Ψ <> PS ;\r
ψ <> ps ;\r
\r
ω <> o $under ;\r
Ω <>  O $under;\r
\r
# at begining or end of word, convert mp to b\r
\r
[^[:L:]$accent] { μπ > b ;\r
μπ } [^[:L:]$accent] > b ;\r
[^[:L:]$accent] { [Μμ][Ππ] > B ;\r
[Μμ][Ππ] } [^[:L:]$accent] > B ;\r
\r
μπ < b ;\r
Μπ < B } $beforeLower ;\r
ΜΠ < B ;\r
\r
# handle diphthongs ending with upsilon\r
\r
ου <> ou ;\r
ΟΥ <> OU ;\r
Ου <> Ou ;\r
οΥ <> oU ;\r
\r
$fmaker = [aeiAEI] $under ? ;\r
$shiftForwardVowels = [[:Mn:]-[\u0308]]; # note: a diaeresis keeps the items separate\r
\r
$fmaker { υ ( $shiftForwardVowels )* } $softener > $1 v $under ;\r
υ $1 < ( $shiftForwardVowels )* v $under ;\r
\r
$fmaker { υ ( $shiftForwardVowels )* } > $1 f $under;\r
υ $1 < ( $shiftForwardVowels )* f $under ;\r
\r
$fmaker { Υ } $softener <> V $under ;\r
$fmaker { Υ <> U $under ;\r
\r
υ <> y ;\r
Υ <> Y ;\r
\r
# NORMAL\r
\r
α <> a ;\r
Α <> A ;\r
\r
β <> v ;\r
Β <> V ;\r
\r
γ } $gammaLike <> n } $egammaLike ;\r
γ <> g ;\r
Γ } $gammaLike <> N } $egammaLike ;\r
Γ <> G ;\r
\r
δ <> d ;\r
Δ <> D ;\r
\r
ε <> e ;\r
Ε <> E ;\r
\r
ζ <> z ;\r
Ζ <> Z ;\r
\r
θ <> th ;\r
Θ } $beforeLower <> Th ;\r
Θ <> TH ;\r
\r
ι <> i ;\r
Ι <> I ;\r
\r
κ <> k ;\r
Κ <> K ;\r
\r
λ <> l ;\r
Λ <> L ;\r
\r
μ <> m ;\r
Μ <> M ;\r
\r
ν } $gammaLike > n\' ;\r
ν <> n ;\r
Ν } $gammaLike <> N\' ;\r
Ν <> N ;\r
\r
ξ <> x ;\r
Ξ <> X ;\r
\r
ο <> o ;\r
Ο <> O ;\r
\r
π <> p ;\r
Π <> P ;\r
\r
ρ <> r ;\r
Ρ <> R ;\r
\r
# insert separator before things that turn into s\r
[Pp] { } [ςσΣϷϸϺϻ] > \' ; \r
\r
# special S variants\r
\r
Ϸ <> Š ; # Ϸ GREEK CAPITAL LETTER SHO Uppercase_Letter Grek - L \r
ϸ <> š ; #ϸ GREEK SMALL LETTER SHO Lowercase_Letter Grek - L \r
Ϻ <> Ŝ ; # Ϻ GREEK CAPITAL LETTER SAN Uppercase_Letter Grek - L \r
ϻ <> ŝ ; # ϻ GREEK SMALL LETTER SAN Lowercase_Letter Grek - L \r
\r
# Caron means exception\r
\r
# before a letter, initial\r
ς } $beforeLetter <> s $under } $beforeLetter;\r
σ } $beforeLetter <> s } $beforeLetter;\r
\r
# otherwise, after a letter = final\r
$afterLetter { σ <> $afterLetter { s $under;\r
$afterLetter { ς <> $afterLetter { s ;\r
\r
# otherwise (isolated) = initial\r
ς <> s $under;\r
σ <> s ;\r
\r
# [Pp] { Σ <> \'S ;\r
Σ <> S ;\r
\r
τ <> t ;\r
Τ <> T ;\r
\r
φ <> f ;\r
Φ <> F ;\r
\r
χ <> ch ;\r
Χ } $beforeLower <> Ch ;\r
Χ <> CH ;\r
\r
# Completeness for ASCII\r
\r
# $ignore = [[:Mark:]''] * ;\r
\r
| ch < h ;\r
| k  < c ;\r
| i  < j ;\r
| k < q ;\r
| b < u } $vowel ;\r
| b < w } $vowel ;\r
| y < u ;\r
| y < w ;\r
\r
| Ch < H ;\r
| K < C ;\r
| I < J ;\r
| K < Q ;\r
| B < W } $vowel ;\r
| B < U } $vowel ;\r
| Y < W ;\r
| Y < U ;\r
\r
# Completeness for Greek\r
\r
ϐ > | β ;\r
ϑ > | θ ;\r
ϒ > | Υ ;\r
ϕ > | φ ;\r
ϖ > | π ;\r
\r
ϰ > | κ ;\r
ϱ > | ρ ;\r
ϲ > | σ ;\r
Ϲ > | Σ; #U+03F9 GREEK CAPITAL LUNATE SIGMA SYMBOL\r
ϳ > j ;\r
ϴ > | Θ ;\r
ϵ > | ε ;\r
µ > | μ ;\r
\r
# delete any trailing ' marks used for roundtripping\r
\r
 < [Ππ] { \' } [Ss] ;\r
 < [Νν] { \' } $egammaLike ;\r
\r
::NFC (NFD) ;\r
\r
# MINIMAL FILTER GENERATED FOR: Latin-Greek/UNGEGN BACKWARD\r
:: ([[[:Latin:][:Mn:][:Me:]] ['\:?]]) ;\r