ICU-3170 fixed the roundtrip test and two greek transliterators.
Excluded some characters; added rules for others. X-SVN-Rev: 14994
This commit is contained in:
Mark Davis
parent
6977909227
commit
b8870b2691
@ -160,16 +160,21 @@ public class RoundTripTest extends TestFmwk {
|
||||
|
||||
String getGreekSet() {
|
||||
// Time bomb
|
||||
return isICU28() ?
|
||||
"[[\u003B\u00B7[:Greek:]-[\u03D7-\u03EF]]&[:Age=3.2:]]" :
|
||||
"[\u003B\u00B7[:Greek:]-[\u03D7-\u03EF]]";
|
||||
return
|
||||
// isICU28() ? "[[\u003B\u00B7[:Greek:]-[\u03D7-\u03EF]]&[:Age=3.2:]]" :
|
||||
"[\u003B\u00B7[[:Greek:]&[:Letter:]]-[" +
|
||||
"\u1D26-\u1D2A" + // L& [5] GREEK LETTER SMALL CAPITAL GAMMA..GREEK LETTER SMALL CAPITAL PSI
|
||||
"\u1D5D-\u1D61" + // Lm [5] MODIFIER LETTER SMALL BETA..MODIFIER LETTER SMALL CHI
|
||||
"\u1D66-\u1D6A" + // L& [5] GREEK SUBSCRIPT SMALL LETTER BETA..GREEK SUBSCRIPT SMALL LETTER CHI
|
||||
"\u03D7-\u03EF" + // \N{GREEK KAI SYMBOL}..\N{COPTIC SMALL LETTER DEI}
|
||||
"]]";
|
||||
}
|
||||
|
||||
|
||||
public void TestGreek() throws IOException, ParseException {
|
||||
long start = System.currentTimeMillis();
|
||||
new Test("Latin-Greek", 50)
|
||||
.test("[a-zA-Z]", getGreekSet(),
|
||||
"[\u00B5\u037A\u03D0-\u03F5]", /* roundtrip exclusions */
|
||||
"[\u00B5\u037A\u03D0-\u03F5\u03F9]", /* roundtrip exclusions */
|
||||
this, new LegalGreek(true));
|
||||
showElapsed(start, "TestGreek");
|
||||
}
|
||||
|
||||
@ -3,14 +3,14 @@
|
||||
# Corporation and others. All Rights Reserved.
|
||||
#--------------------------------------------------------------------
|
||||
# $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/impl/data/Attic/Transliterator_Greek_Latin.txt,v $
|
||||
# $Date: 2002/07/21 08:39:23 $
|
||||
# $Revision: 1.21 $
|
||||
# $Date: 2004/04/16 14:16:48 $
|
||||
# $Revision: 1.22 $
|
||||
#--------------------------------------------------------------------
|
||||
|
||||
# Rules are predicated on running NFD first, and NFC afterwards
|
||||
# :: [\u0000-\u007F \u0370-\u03FF [:Greek:] [:nonspacing mark:]] ;
|
||||
# MINIMAL FILTER GENERATED FOR: Greek-Latin
|
||||
:: [;\u00B5\u00B7\u00C4\u00CB\u00CF\u00D6\u00DC\u00E4\u00EB\u00EF\u00F6\u00FC\u00FF-\u0101\u0112-\u0113\u012A-\u012B\u014C-\u014D\u016A-\u016B\u0178\u01D5-\u01DC\u01DE-\u01E3\u01EC-\u01ED\u022A-\u022D\u0230-\u0233\u0304\u0308\u0313-\u0314\u0342-\u0345\u037A\u037E\u0386-\u038A\u038C\u038E-\u03A1\u03A3-\u03CE\u03D0-\u03D7\u03DB\u03DD\u03DF\u03E1\u03E3\u03E5\u03E7\u03E9\u03EB\u03ED\u03EF-\u03F5\u0401\u0407\u0451\u0457\u04D2-\u04D3\u04DA-\u04DF\u04E2-\u04E7\u04EA-\u04F1\u04F4-\u04F5\u04F8-\u04F9\u1E14-\u1E17\u1E20-\u1E21\u1E26-\u1E27\u1E2E-\u1E2F\u1E38-\u1E39\u1E4E-\u1E53\u1E5C-\u1E5D\u1E7A-\u1E7B\u1E84-\u1E85\u1E8C-\u1E8D\u1E97\u1F00-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC1-\u1FC4\u1FC6-\u1FCC\u1FCF-\u1FD3\u1FD6-\u1FDB\u1FDF-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u2126] ;
|
||||
:: [;\u00B5\u00B7\u00C4\u00CB\u00CF\u00D6\u00DC\u00E4\u00EB\u00EF\u00F6\u00FC\u00FF-\u0101\u0112-\u0113\u012A-\u012B\u014C-\u014D\u016A-\u016B\u0178\u01D5-\u01DC\u01DE-\u01E3\u01EC-\u01ED\u022A-\u022D\u0230-\u0233\u0304\u0308\u0313-\u0314\u0342-\u0345\u037A\u037E\u0386-\u038A\u038C\u038E-\u03A1\u03A3-\u03CE\u03D0-\u03D7\u03DB\u03DD\u03DF\u03E1\u03E3\u03E5\u03E7\u03E9\u03EB\u03ED\u03EF-\u03F5\u03F7-\u07FB\u0401\u0407\u0451\u0457\u04D2-\u04D3\u04DA-\u04DF\u04E2-\u04E7\u04EA-\u04F1\u04F4-\u04F5\u04F8-\u04F9\u1E14-\u1E17\u1E20-\u1E21\u1E26-\u1E27\u1E2E-\u1E2F\u1E38-\u1E39\u1E4E-\u1E53\u1E5C-\u1E5D\u1E7A-\u1E7B\u1E84-\u1E85\u1E8C-\u1E8D\u1E97\u1F00-\u1F15\u1F18-\u1F1D\u1F20-\u1F45\u1F48-\u1F4D\u1F50-\u1F57\u1F59\u1F5B\u1F5D\u1F5F-\u1F7D\u1F80-\u1FB4\u1FB6-\u1FBC\u1FBE\u1FC1-\u1FC4\u1FC6-\u1FCC\u1FCF-\u1FD3\u1FD6-\u1FDB\u1FDF-\u1FEC\u1FF2-\u1FF4\u1FF6-\u1FFC\u2126\u03F9] ;
|
||||
|
||||
:: NFD (NFC) ;
|
||||
|
||||
@ -251,10 +251,16 @@ $smooth > ;
|
||||
ρ <> r ;
|
||||
Ρ <> R ;
|
||||
|
||||
# insert separator
|
||||
# insert separator before things that turn into s
|
||||
|
||||
[Pp] { } ς > \' ;
|
||||
[Pp] { } σ > \' ;
|
||||
[Pp] { } [ςσΣϷϸϺϻ] > \' ;
|
||||
|
||||
# special S variants
|
||||
|
||||
Ϸ <> Š ; # Ϸ GREEK CAPITAL LETTER SHO Uppercase_Letter Grek - L
|
||||
ϸ <> š ; #ϸ GREEK SMALL LETTER SHO Lowercase_Letter Grek - L
|
||||
Ϻ <> Ŝ ; # Ϻ GREEK CAPITAL LETTER SAN Uppercase_Letter Grek - L
|
||||
ϻ <> ŝ ; # ϻ GREEK SMALL LETTER SAN Lowercase_Letter Grek - L
|
||||
|
||||
# underbar means exception
|
||||
|
||||
@ -270,7 +276,7 @@ $afterLetter { ς <> $afterLetter { s ;
|
||||
ς <> s $underbar;
|
||||
σ <> s ;
|
||||
|
||||
[Pp] { Σ <> \'S ;
|
||||
# [Pp] { Σ <> \'S ;
|
||||
Σ <> S ;
|
||||
|
||||
τ <> t ;
|
||||
@ -322,6 +328,7 @@ $rough <> h ;
|
||||
ϰ > | κ ;
|
||||
ϱ > | ρ ;
|
||||
ϲ > | σ ;
|
||||
Ϲ > | Σ; #U+03F9 GREEK CAPITAL LUNATE SIGMA SYMBOL
|
||||
ϳ > j ;
|
||||
ϴ > | Θ ;
|
||||
ϵ > | ε ;
|
||||
|
||||
@ -3,15 +3,15 @@
|
||||
# Corporation and others. All Rights Reserved.
|
||||
#--------------------------------------------------------------------
|
||||
# $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/impl/data/Attic/Transliterator_Greek_Latin_UNGEGN.txt,v $
|
||||
# $Date: 2002/07/26 16:09:04 $
|
||||
# $Revision: 1.2 $
|
||||
# $Date: 2004/04/16 14:16:47 $
|
||||
# $Revision: 1.3 $
|
||||
#--------------------------------------------------------------------
|
||||
# For modern Greek, based on UNGEGN rules.
|
||||
|
||||
# Rules are predicated on running NFD first, and NFC afterwards
|
||||
# MINIMAL FILTER GENERATED FOR: Greek-Latin/UNGEGN
|
||||
# WARNING: need to add accents to both filters ###
|
||||
# :: [́̄̆̈;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] ;
|
||||
# :: [́̄̆̈;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] ;
|
||||
|
||||
:: [[[:Greek:][:Mn:][:Me:]] [\:-;?\u00B7\u037E\u0387]] ;
|
||||
::NFD (NFC) ;
|
||||
@ -169,8 +169,15 @@ $fmaker { Υ <> U $under ;
|
||||
ρ <> r ;
|
||||
Ρ <> R ;
|
||||
|
||||
[Pp] { } ς > \' ;
|
||||
[Pp] { } σ > \' ;
|
||||
# insert separator before things that turn into s
|
||||
[Pp] { } [ςσΣϷϸϺϻ] > \' ;
|
||||
|
||||
# special S variants
|
||||
|
||||
Ϸ <> Š ; # Ϸ GREEK CAPITAL LETTER SHO Uppercase_Letter Grek - L
|
||||
ϸ <> š ; #ϸ GREEK SMALL LETTER SHO Lowercase_Letter Grek - L
|
||||
Ϻ <> Ŝ ; # Ϻ GREEK CAPITAL LETTER SAN Uppercase_Letter Grek - L
|
||||
ϻ <> ŝ ; # ϻ GREEK SMALL LETTER SAN Lowercase_Letter Grek - L
|
||||
|
||||
# Caron means exception
|
||||
|
||||
@ -186,7 +193,7 @@ $afterLetter { ς <> $afterLetter { s ;
|
||||
ς <> s $under;
|
||||
σ <> s ;
|
||||
|
||||
[Pp] { Σ <> \'S ;
|
||||
# [Pp] { Σ <> \'S ;
|
||||
Σ <> S ;
|
||||
|
||||
τ <> t ;
|
||||
@ -232,6 +239,7 @@ $afterLetter { ς <> $afterLetter { s ;
|
||||
ϰ > | κ ;
|
||||
ϱ > | ρ ;
|
||||
ϲ > | σ ;
|
||||
Ϲ > | Σ; #U+03F9 GREEK CAPITAL LUNATE SIGMA SYMBOL
|
||||
ϳ > j ;
|
||||
ϴ > | Θ ;
|
||||
ϵ > | ε ;
|
||||
|
||||
Loading…
Reference in New Issue
Block a user