diff options
Diffstat (limited to 'Build/source/libs/teckit/TECkit-2.5.3/test/SILGreek2004-04-27.map')
-rw-r--r-- | Build/source/libs/teckit/TECkit-2.5.3/test/SILGreek2004-04-27.map | 697 |
1 files changed, 697 insertions, 0 deletions
diff --git a/Build/source/libs/teckit/TECkit-2.5.3/test/SILGreek2004-04-27.map b/Build/source/libs/teckit/TECkit-2.5.3/test/SILGreek2004-04-27.map new file mode 100644 index 00000000000..11a80cfbb47 --- /dev/null +++ b/Build/source/libs/teckit/TECkit-2.5.3/test/SILGreek2004-04-27.map @@ -0,0 +1,697 @@ +;SILGreek.map +;Copyright (c) 2002-2011 SIL International. + +; SIL Greek + +EncodingName "SIL-GREEK_GALATIA-2001" +DescriptiveName "SIL Greek Encoding (Galatia)" + +Version "1.0b5" +Contact "mailto:nrsi@sil.org" +RegistrationAuthority "SIL International" +RegistrationName "Greek (Galatia)" + +LHSFlags () +RHSFlags (ExpectsNFD) + +; 1.0b4 2004-04-26 JK added pass to map between Galatia and Unicode order for marks +; 1.0b5 2004-04-27 JK added mappings for 150,151,160,164,172,182 in Display encoding + +; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * + +Pass(Byte) + +; First we map precomposed "display" forms to their equivalent "basic" sequences +; while still in the Byte (SIL Greek legacy encoding) world, except that we +; maintain the final/non-final sigma distinction + +Class [LTR] = ( 'a'..'u' 'w'..'z' 'A'..'U' 'W'..'Z' '`^_@"' "'" \ + 128..149 152..159 161..163 165..171 173..181 184..255) + +; make sigma into final form if not followed by a letter +'s' / _ ^[LTR] > 'v' + +;This is copied directly from GRCO-BA.CCT - Greek Composite to Basic conversion +;and then the unidirectional '>' operators changed to bidirectional '<>' +;'v' > 's' ; leave final sigma alone, as Unicode has separate codes + +0xCF <> 'Hr' +0xBF <> 'hr' + +0xAD 'A' <> 'HA' "'" +0xAE 'A' <> 'HA`' +0xAF 'A' <> 'HA^' +0xA9 'A' <> 'hA' "'" +0xAA 'A' <> 'hA`' +0xAB 'A' <> 'hA^' + +0xAD 'E' <> 'HE' "'" +0xA9 'E' <> 'hE' "'" +0xAA 'E' <> 'hE`' + +0xAD 'J' <> 'HJ' "'" +0xAE 'J' <> 'HJ`' +0xAF 'J' <> 'HJ^' +0xA9 'J' <> 'hJ' "'" +0xAA 'J' <> 'hJ`' +0xAB 'J' <> 'hJ^' + +0xAD 'I' <> 'HI' "'" +0xAF 'I' <> 'HI^' +0xA9 'I' <> 'hI' "'" +0xAA 'I' <> 'hI`' +0xAB 'I' <> 'hI^' + +0xAD 'O' <> 'HO' "'" +0xAE 'O' <> 'HO`' +0xA9 'O' <> 'hO' "'" +0xAA 'O' <> 'hO`' + +0xAD 'U' <> 'HU' "'" +0xAE 'U' <> 'HU`' +0xAF 'U' <> 'HU^' +0xA9 'U' <> 'hU' "'" +0xAA 'U' <> 'hU`' +0xAB 'U' <> 'hU^' + +0xAD 'W' <> 'HW' "'" +0xAE 'W' <> 'HW`' +0xAF 'W' <> 'HW^' +0xA9 'W' <> 'hW' "'" +0xAA 'W' <> 'hW`' +0xAB 'W' <> 'hW^' + +0x87 <> 'Ha' +0x80 <> 'a' "'" +0x88 <> 'Ha' "'" +0x81 <> 'a`' +0x89 <> 'Ha`' +0x82 <> 'a^' +0x8A <> 'Ha^' +0x83 <> 'ha' +0x84 <> 'ha' "'" +0x85 <> 'ha`' +0x86 <> 'ha^' + +0x9D <> 'He' +0x99 <> 'e' "'" +0x9E <> 'He' "'" +0x9A <> 'e`' +0x9B <> 'he' +0x9C <> 'he' "'" +0x9F <> 'he`' + +0xD7 <> 'Hj' +0xD0 <> 'j' "'" +0xD8 <> 'Hj' "'" +0xD1 <> 'j`' +0xD9 <> 'Hj`' +0xD2 <> 'j^' +0xDA <> 'Hj^' +0xD3 <> 'hj' +0xD4 <> 'hj' "'" +0xD5 <> 'hj`' +0xD6 <> 'hj^' + +0xB8 <> 'Hi' +0xB0 <> 'i' "'" +0xB9 <> 'Hi' "'" +0xB1 <> 'i`' +0xB2 <> 'i^' +0xBA <> 'Hi^' +0xB3 <> 'hi' +0xB4 <> 'hi' "'" +0xBE <> 'hi`' +0xB5 <> 'hi^' + +0xEC <> 'Ho' +0xE7 <> 'o' "'" +0xED <> 'Ho' "'" +0xE8 <> 'o`' +0xCE <> 'Ho`' +0xE9 <> 'ho' +0xEA <> 'ho' "'" +0xEB <> 'ho`' + +0xC7 <> 'Hu' +0xC0 <> 'u' "'" +0xC8 <> 'Hu' "'" +0xC1 <> 'u`' +0xC9 <> 'Hu`' +0xC2 <> 'u^' +0xCA <> 'Hu^' +0xC3 <> 'hu' +0xC4 <> 'hu' "'" +0xC5 <> 'hu`' +0xC6 <> 'hu^' + +0xF7 <> 'Hw' +0xF0 <> 'w' "'" +0xF8 <> 'Hw' "'" +0xF1 <> 'w`' +0xF9 <> 'Hw`' +0xF2 <> 'w^' +0xFA <> 'Hw^' +0xF3 <> 'hw' +0xF4 <> 'hw' "'" +0xF5 <> 'hw`' +0xF6 <> 'hw^' + +0x8B <> 'a_' +0x93 <> 'Ha_' +0x8C <> 'a_' "'" +0x94 <> 'Ha_' "'" +0x8D <> 'a_`' +0x95 <> 'Ha_`' +0x8E <> 'a_^' +0x98 <> 'Ha_^' +0x8F <> 'ha_' +0x90 <> 'ha_' "'" +0x91 <> 'ha_`' +0x92 <> 'ha_^' + +0xDB <> 'j_' +0xE3 <> 'Hj_' +0xDC <> 'j_' "'" +0xE4 <> 'Hj_' "'" +0xDD <> 'j_`' +0xE5 <> 'Hj_`' +0xDE <> 'j_^' +0xE6 <> 'Hj_^' +0xDF <> 'hj_' +0xE0 <> 'hj_' "'" +0xE1 <> 'hj_`' +0xE2 <> 'hj_^' + +0xFB <> 'w_' +0xA5 <> 'Hw_' +0xFC <> 'w_' "'" +0xA6 <> 'Hw_' "'" +0xFD <> 'w_`' +0xA7 <> 'Hw_`' +0xFE <> 'w_^' +0xA8 <> 'Hw_^' +0xFF <> 'hw_' +0xA1 <> 'hw_' "'" +0xA2 <> 'hw_`' +0xA3 <> 'hw_^' + +0xBB <> 'i"' +0xBC <> 'i"' "'" +0xBD <> 'i"`' + +0xCB <> 'u"' +0xCC <> 'u"' "'" +0xCD <> 'u"`' + +'A' 0xB8 <> 'HAi' +'A' 0xB9 <> 'HAi' "'" +'A' 0xBA <> 'HAi^' +'A' 0xB3 <> 'hAi' +'A' 0xB4 <> 'hAi' "'" +'A' 0xBE <> 'hAi`' +'A' 0xB5 <> 'hAi^' + +'a' 0xB8 <> 'Hai' +'a' 0xB0 <> 'ai' "'" +'a' 0xB9 <> 'Hai' "'" +'a' 0xB1 <> 'ai`' +'a' 0xB2 <> 'ai^' +'a' 0xBA <> 'Hai^' +'a' 0xB3 <> 'hai' +'a' 0xB4 <> 'hai' "'" +'a' 0xBE <> 'hai`' +'a' 0xB5 <> 'hai^' + +'E' 0xB8 <> 'HEi' +'E' 0xB9 <> 'HEi' "'" +'E' 0xBA <> 'HEi^' +'E' 0xB3 <> 'hEi' +'E' 0xB4 <> 'hEi' "'" +'E' 0xBE <> 'hEi`' +'E' 0xB5 <> 'hEi^' + +'e' 0xB8 <> 'Hei' +'e' 0xB0 <> 'ei' "'" +'e' 0xB9 <> 'Hei' "'" +'e' 0xB1 <> 'ei`' +'e' 0xB2 <> 'ei^' +'e' 0xBA <> 'Hei^' +'e' 0xB3 <> 'hei' +'e' 0xB4 <> 'hei' "'" +'e' 0xBE <> 'hei`' +'e' 0xB5 <> 'hei^' + +'O' 0xB8 <> 'HOi' +'O' 0xB9 <> 'HOi' "'" +'O' 0xBA <> 'HOi^' +'O' 0xB3 <> 'hOi' +'O' 0xB4 <> 'hOi' "'" +'O' 0xBE <> 'hOi`' +'O' 0xB5 <> 'hOi^' + +'o' 0xB8 <> 'Hoi' +'o' 0xB0 <> 'oi' "'" +'o' 0xB9 <> 'Hoi' "'" +'o' 0xB1 <> 'oi`' +'o' 0xB2 <> 'oi^' +'o' 0xBA <> 'Hoi^' +'o' 0xB3 <> 'hoi' +'o' 0xB4 <> 'hoi' "'" +'o' 0xBE <> 'hoi`' +'o' 0xB5 <> 'hoi^' + +'U' 0xB8 <> 'HUi' +'U' 0xB9 <> 'HUi' "'" +'U' 0xBA <> 'HUi^' +'U' 0xB3 <> 'hUi' +'U' 0xB4 <> 'hUi' "'" +'U' 0xBE <> 'hUi`' +'U' 0xB5 <> 'hUi^' + +'u' 0xB8 <> 'Hui' +'u' 0xB0 <> 'ui' "'" +'u' 0xB9 <> 'Hui' "'" +'u' 0xB1 <> 'ui`' +'u' 0xB2 <> 'ui^' +'u' 0xBA <> 'Hui^' +'u' 0xB3 <> 'hui' +'u' 0xB4 <> 'hui' "'" +'u' 0xBE <> 'hui`' +'u' 0xB5 <> 'hui^' + +'A' 0xC7 <> 'HAu' +'A' 0xC8 <> 'HAu' "'" +'A' 0xC9 <> 'HAu`' +'A' 0xCA <> 'HAu^' +'A' 0xC3 <> 'hAu' +'A' 0xC4 <> 'hAu' "'" +'A' 0xC5 <> 'hAu`' +'A' 0xC6 <> 'hAu^' + +'a' 0xC7 <> 'Hau' +'a' 0xC0 <> 'au' "'" +'a' 0xC8 <> 'Hau' "'" +'a' 0xC1 <> 'au`' +'a' 0xC9 <> 'Hau`' +'a' 0xC2 <> 'au^' +'a' 0xCA <> 'Hau^' +'a' 0xC3 <> 'hau' +'a' 0xC4 <> 'hau' "'" +'a' 0xC5 <> 'hau`' +'a' 0xC6 <> 'hau^' + +'E' 0xC7 <> 'HEu' +'E' 0xC8 <> 'HEu' "'" +'E' 0xC9 <> 'HEu`' +'E' 0xCA <> 'HEu^' +'E' 0xC3 <> 'hEu' +'E' 0xC4 <> 'hEu' "'" +'E' 0xC5 <> 'hEu`' +'E' 0xC6 <> 'hEu^' + +'e' 0xC7 <> 'Heu' +'e' 0xC0 <> 'eu' "'" +'e' 0xC8 <> 'Heu' "'" +'e' 0xC1 <> 'eu`' +'e' 0xC9 <> 'Heu`' +'e' 0xC2 <> 'eu^' +'e' 0xCA <> 'Heu^' +'e' 0xC3 <> 'heu' +'e' 0xC4 <> 'heu' "'" +'e' 0xC5 <> 'heu`' +'e' 0xC6 <> 'heu^' + +'J' 0xC7 <> 'HJu' +'J' 0xC8 <> 'HJu' "'" +'J' 0xC9 <> 'HJu`' +'J' 0xCA <> 'HJu^' +'J' 0xC3 <> 'hJu' +'J' 0xC4 <> 'hJu' "'" +'J' 0xC5 <> 'hJu`' +'J' 0xC6 <> 'hJu^' + +'j' 0xC7 <> 'Hju' +'j' 0xC0 <> 'ju' "'" +'j' 0xC8 <> 'Hju' "'" +'j' 0xC1 <> 'ju`' +'j' 0xC9 <> 'Hju`' +'j' 0xC2 <> 'ju^' +'j' 0xCA <> 'Hju^' +'j' 0xC3 <> 'hju' +'j' 0xC4 <> 'hju' "'" +'j' 0xC5 <> 'hju`' +'j' 0xC6 <> 'hju^' + +'O' 0xC7 <> 'HOu' +'O' 0xC8 <> 'HOu' "'" +'O' 0xC9 <> 'HOu`' +'O' 0xCA <> 'HOu^' +'O' 0xC3 <> 'hOu' +'O' 0xC4 <> 'hOu' "'" +'O' 0xC5 <> 'hOu`' +'O' 0xC6 <> 'hOu^' + +'o' 0xC7 <> 'Hou' +'o' 0xC0 <> 'ou' "'" +'o' 0xC8 <> 'Hou' "'" +'o' 0xC1 <> 'ou`' +'o' 0xC9 <> 'Hou`' +'o' 0xC2 <> 'ou^' +'o' 0xCA <> 'Hou^' +'o' 0xC3 <> 'hou' +'o' 0xC4 <> 'hou' "'" +'o' 0xC5 <> 'hou`' +'o' 0xC6 <> 'hou^' + +'HA' 0xBB <> 'HAi"' +'HA' 0xBC <> 'HAi"' "'" +'HA' 0xBD <> 'HAi"`' +'hA' 0xBB <> 'hAi"' +'hA' 0xBC <> 'hAi"' "'" +'hA' 0xBD <> 'hAi"`' +'a' 0xBB <> 'ai"' +0x87 0xBB <> 'Hai"' +'a' 0xBC <> 'ai"' "'" +0x87 0xBC <> 'Hai"' "'" +'a' 0xBD <> 'ai"`' +0x87 0xBD <> 'Hai"`' +0x83 0xBB <> 'hai"' +0x83 0xBC <> 'hai"' "'" +0x83 0xBD <> 'hai"`' + +'HE' 0xBB <> 'HEi"' +'HE' 0xBC <> 'HEi"' "'" +'HE' 0xBD <> 'HEi"`' +'hE' 0xBB <> 'hEi"' +'hE' 0xBC <> 'hEi"' "'" +'hE' 0xBD <> 'hEi"`' +'e' 0xBB <> 'ei"' +0x9D 0xBB <> 'Hei"' +'e' 0xBC <> 'ei"' "'" +0x9D 0xBC <> 'Hei"' "'" +'e' 0xBD <> 'ei"`' +0x9D 0xBD <> 'Hei"`' +0x9B 0xBB <> 'hei"' +0x9B 0xBC <> 'hei"' "'" +0x9B 0xBD <> 'hei"`' + +'HO' 0xBB <> 'HOi"' +'HO' 0xBC <> 'HOi"' "'" +'HO' 0xBD <> 'HOi"`' +'hO' 0xBB <> 'hOi"' +'hO' 0xBC <> 'hOi"' "'" +'hO' 0xBD <> 'hOi"`' +'o' 0xBB <> 'oi"' +0xEC 0xBB <> 'Hoi"' +'o' 0xBC <> 'oi"' "'" +0xEC 0xBC <> 'Hoi"' "'" +'o' 0xBD <> 'oi"`' +0xEC 0xBD <> 'Hoi"`' +0xE9 0xBB <> 'hoi"' +0xE9 0xBC <> 'hoi"' "'" +0xE9 0xBD <> 'hoi"`' + +'HU' 0xBB <> 'HUi"' +'HU' 0xBC <> 'HUi"' "'" +'HU' 0xBD <> 'HUi"`' +'hU' 0xBB <> 'hUi"' +'hU' 0xBC <> 'hUi"' "'" +'hU' 0xBD <> 'hUi"`' +'u' 0xBB <> 'ui"' +0xC7 0xBB <> 'Hui"' +'u' 0xBC <> 'ui"' "'" +0xC7 0xBC <> 'Hui"' "'" +'u' 0xBD <> 'ui"`' +0xC7 0xBD <> 'Hui"`' +0xC3 0xBB <> 'hui"' +0xC3 0xBC <> 'hui"' "'" +0xC3 0xBD <> 'hui"`' + +'HA' 0xCB <> 'HAu"' +'HA' 0xCC <> 'HAu"' "'" +'HA' 0xCD <> 'HAu"`' +'hA' 0xCB <> 'hAu"' +'hA' 0xCC <> 'hAu"' "'" +'hA' 0xCD <> 'hAu"`' +'a' 0xCB <> 'au"' +0x87 0xCB <> 'Hau"' +'a' 0xCC <> 'au"' "'" +0x87 0xCC <> 'Hau"' "'" +'a' 0xCD <> 'au"`' +0x87 0xCD <> 'Hau"`' +0x83 0xCB <> 'hau"' +0x83 0xCC <> 'hau"' "'" +0x83 0xCD <> 'hau"`' + +'HE' 0xCB <> 'HEu"' +'HE' 0xCC <> 'HEu"' "'" +'HE' 0xCD <> 'HEu"`' +'hE' 0xCB <> 'hEu"' +'hE' 0xCC <> 'hEu"' "'" +'hE' 0xCD <> 'hEu"`' +'e' 0xCB <> 'eu"' +0x9D 0xCB <> 'Heu"' +'e' 0xCC <> 'eu"' "'" +0x9D 0xCC <> 'Heu"' "'" +'e' 0xCD <> 'eu"`' +0x9D 0xCD <> 'Heu"`' +0x9B 0xCB <> 'heu"' +0x9B 0xCC <> 'heu"' "'" +0x9B 0xCD <> 'heu"`' + +'HJ' 0xCB <> 'HJu"' +'HJ' 0xCC <> 'HJu"' "'" +'HJ' 0xCD <> 'HJu"`' +'hJ' 0xCB <> 'hJu"' +'hJ' 0xCC <> 'hJu"' "'" +'hJ' 0xCD <> 'hJu"`' +'j' 0xCB <> 'ju"' +0xD7 0xCB <> 'Hju"' +'j' 0xCC <> 'ju"' "'" +0xD7 0xCC <> 'Hju"' "'" +'j' 0xCD <> 'ju"`' +0xD7 0xCD <> 'Hju"`' +0xD3 0xCB <> 'hju"' +0xD3 0xCC <> 'hju"' "'" +0xD3 0xCD <> 'hju"`' + +'HO' 0xCB <> 'HOu"' +'HO' 0xCC <> 'HOu"' "'" +'HO' 0xCD <> 'HOu"`' +'hO' 0xCB <> 'hOu"' +'hO' 0xCC <> 'hOu"' "'" +'hO' 0xCD <> 'hOu"`' +'o' 0xCB <> 'ou"' +0xEC 0xCB <> 'Hou"' +'o' 0xCC <> 'ou"' "'" +0xEC 0xCC <> 'Hou"' "'" +'o' 0xCD <> 'ou"`' +0xEC 0xCD <> 'Hou"`' +0xE9 0xCB <> 'hou"' +0xE9 0xCC <> 'hou"' "'" +0xE9 0xCD <> 'hou"`' + +0xAD <> 'H' "'" +0xAE <> 'H`' +0xAF <> 'H^' +0xA9 <> 'h' "'" +0xAA <> 'h`' +0xAB <> 'h^' + +0xEE <> '"' "'" +0xEF <> '"`' + +; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * + +Pass(Byte) +; shuffle marks, as Unicode canonical order does not correspond to Galatia order + +Class[dia] = ('`' "'" '^' '"') + +'_' ([dia]+)=dias <> @dias '_' + +; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * + +Pass(Byte_Unicode) + +; Now make the jump from byte to Unicode space. +; Here, we deal only with the decomposed forms, not precomposed. + +ByteDefault 183 ; 183 is "bullet" in the SIL Greek Display encoding + ; (note that this is actually a valid character code, + ; though probably not often used in legacy Greek text) +UniDefault replacement_character + +; there are separate namespaces for Byte and Unicode classes, +; allowing us to use the same name for classes with corresponding content + +ByteClass [CTL] = ( 0x00 .. 0x1f 0x7f ) +UniClass [CTL] = ( U+0000 .. U+001f U+007f ) + +[CTL] <> [CTL] + +' ' <> space +'!' <> exclamation_mark +'"' <> combining_diaeresis +'#' <> no_break_space +'$' <> left_pointing_double_angle_quotation_mark +'%' <> right_pointing_double_angle_quotation_mark +'&' <> ampersand +"'" <> combining_acute_accent +'(' <> left_parenthesis +')' <> right_parenthesis +'*' <> asterisk +'+' <> plus_sign +',' <> comma +'-' <> hyphen_minus +'.' <> full_stop +'/' <> solidus + +'0' <> digit_zero +'1' <> digit_one +'2' <> digit_two +'3' <> digit_three +'4' <> digit_four +'5' <> digit_five +'6' <> digit_six +'7' <> digit_seven +'8' <> digit_eight +'9' <> digit_nine +':' <> colon +';' < greek_ano_teleia ; greek semicolon +';' <> middle_dot ; canonical decomposition of greek semicolon +'<' <> less_than_sign +'=' <> equals_sign +'>' <> greater_than_sign +'?' < greek_question_mark +'?' <> semicolon ; canonical decomposition of greek question mark + +'@' <> right_single_quotation_mark +'@' < modifier_letter_apostrophe ; mapping of '@' in earlier draft of this file +'A' <> greek_capital_letter_alpha +'B' <> greek_capital_letter_beta +'C' <> greek_capital_letter_chi +'D' <> greek_capital_letter_delta +'E' <> greek_capital_letter_epsilon +'F' <> greek_capital_letter_phi +'G' <> greek_capital_letter_gamma +'H' <> combining_comma_above +'I' <> greek_capital_letter_iota +'J' <> greek_capital_letter_eta +'K' <> greek_capital_letter_kappa +'L' <> greek_capital_letter_lamda +'M' <> greek_capital_letter_mu +'N' <> greek_capital_letter_nu +'O' <> greek_capital_letter_omicron + +'P' <> greek_capital_letter_pi +'Q' <> greek_capital_letter_theta +'R' <> greek_capital_letter_rho +'S' <> greek_capital_letter_sigma +'T' <> greek_capital_letter_tau +'U' <> greek_capital_letter_upsilon +; 'V' ; <unused> +'W' <> greek_capital_letter_omega +'X' <> greek_capital_letter_xi +'Y' <> greek_capital_letter_psi +'Z' <> greek_capital_letter_zeta +'[' <> left_square_bracket +'\' <> reverse_solidus +']' <> right_square_bracket +'^' <> combining_greek_perispomeni ; circumflex +'_' <> combining_greek_ypogegrammeni ; iota subscript + +'`' <> combining_grave_accent +'a' <> greek_small_letter_alpha +'b' <> greek_small_letter_beta +'c' <> greek_small_letter_chi +'d' <> greek_small_letter_delta +'e' <> greek_small_letter_epsilon +'f' <> greek_small_letter_phi +'g' <> greek_small_letter_gamma +'h' <> combining_reversed_comma_above +'i' <> greek_small_letter_iota +'j' <> greek_small_letter_eta +'k' <> greek_small_letter_kappa +'l' <> greek_small_letter_lamda +'m' <> greek_small_letter_mu +'n' <> greek_small_letter_nu +'o' <> greek_small_letter_omicron + +'p' <> greek_small_letter_pi +'q' <> greek_small_letter_theta +'r' <> greek_small_letter_rho +'s' <> greek_small_letter_sigma +'t' <> greek_small_letter_tau +'u' <> greek_small_letter_upsilon +'v' <> greek_small_letter_final_sigma +'w' <> greek_small_letter_omega +'x' <> greek_small_letter_xi +'y' <> greek_small_letter_psi +'z' <> greek_small_letter_zeta +'{' <> left_curly_bracket + +'|' <> vertical_line +'|b' <> greek_beta_symbol ; curly beta +'|f' <> greek_small_letter_digamma ; digamma +'|G' <> greek_letter_digamma ; Digamma +'|g' > greek_small_letter_digamma ; digamma (which SIL forms should have priority?) +'|w' <> greek_pi_symbol ; omega pi +'|q' <> greek_small_letter_koppa ; qoppa +'|Q' <> greek_letter_koppa ; Qoppa +'|R' > greek_letter_koppa ; Qoppa +'|k' > greek_small_letter_koppa ; qoppa +'|K' > greek_letter_koppa ; Qoppa +'|p' <> greek_small_letter_sampi ; sampi +'|P' <> greek_letter_sampi ; Sampi +'|m' > greek_small_letter_sampi ; sampi +'|M' > greek_letter_sampi ; Sampi +'|s' <> greek_lunate_sigma_symbol ; lunate sigma +'|S' > greek_lunate_sigma_symbol ; lunate sigma uppercase not available in Unicode +'|t' <> greek_small_letter_stigma ; stigma +'|T' <> greek_letter_stigma ; Stigma +'|i' <> turned_greek_small_letter_iota combining_tilde_below ; turned iota tilde below +'|U' <> greek_upsilon_with_hook_symbol ; Upsilon hook +'|h' <> greek_numeral_sign ; upper numeral sign +'|H' <> greek_lower_numeral_sign ; lower numeral sign +'|n' <> zero_width_joiner +'|#' <> zero_width_non_joiner + +'}' <> right_curly_bracket +'~' <> em_dash + +150 <> en_dash +151 > em_dash +160 <> no_break_space +164 <> currency_sign +172 <> not_sign +182 <> pilcrow_sign +183 <> bullet + +; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * + +Pass(Unicode) + +; In Unicode space, reorder breathing/vowel sequences from SIL Basic to Unicode order + +Class [BR] = ( combining_comma_above combining_reversed_comma_above ) +Class [aeo] = ( U+0391 U+0395 U+039f U+03b1 U+03b5 U+03bf ) ; Unicode names are so verbose! +Class [iu] = ( U+0399 U+03a5 U+03b9 U+03c5 ) +Class [j] = ( U+0397 U+03b7 ) +Class [u] = ( U+03a5 U+03c5 ) +Class [i] = ( U+0399 U+03b9 ) +Class [vowelrho] = ( U+0391 U+0395 U+0399 U+039f U+03a5 U+0397 U+03a9 U+03a1 \ + U+03b1 U+03b5 U+03b9 U+03bf U+03c5 U+03b7 U+03c9 U+03c1 ) + +[BR]=b [aeo]=v1 [iu]=v2 / _ combining_diaeresis <> @v1 @b @v2 / _ combining_diaeresis +[BR]=b [aeo]=v1 [iu]=v2 <> @v1 @v2 @b +[BR]=b [j]=v1 [u]=v2 / _ combining_diaeresis <> @v1 @b @v2 / _ combining_diaeresis +[BR]=b [j]=v1 [u]=v2 <> @v1 @v2 @b +[BR]=b [u]=v1 [i]=v2 / _ combining_diaeresis <> @v1 @b @v2 / _ combining_diaeresis +[BR]=b [u]=v1 [i]=v2 <> @v1 @v2 @b +[BR]=b [vowelrho]=v <> @v @b + + +; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * +; (end of file) |