102 lines
4.9 KiB
Ragel
102 lines
4.9 KiB
Ragel
# The following Ragel file was autogenerated with ragel/unicode2ragel.rb
|
|
# from: http://www.unicode.org/Public/8.0.0/ucd/Scripts.txt
|
|
#
|
|
# It defines ["Hangul", "Han", "Hiragana"].
|
|
#
|
|
# To use this, make sure that your alphtype is set to unsigned char,
|
|
# and that your input is in utf8.
|
|
|
|
%%{
|
|
machine SCRIPTS;
|
|
|
|
Hangul =
|
|
0xE1 0x84 0x80..0xFF #Lo [256] HANGUL CHOSEONG KIYEOK..HANGUL...
|
|
| 0xE1 0x85..0x86 0x00..0xFF #
|
|
| 0xE1 0x87 0x00..0xBF #
|
|
| 0xE3 0x80 0xAE..0xAF #Mc [2] HANGUL SINGLE DOT TONE MARK..HANGU...
|
|
| 0xE3 0x84 0xB1..0xFF #Lo [94] HANGUL LETTER KIYEOK..HANGUL L...
|
|
| 0xE3 0x85..0x85 0x00..0xFF #
|
|
| 0xE3 0x86 0x00..0x8E #
|
|
| 0xE3 0x88 0x80..0x9E #So [31] PARENTHESIZED HANGUL KIYEOK..PAREN...
|
|
| 0xE3 0x89 0xA0..0xBE #So [31] CIRCLED HANGUL KIYEOK..CIRCLED HAN...
|
|
| 0xEA 0xA5 0xA0..0xBC #Lo [29] HANGUL CHOSEONG TIKEUT-MIEUM..HANG...
|
|
| 0xEA 0xB0 0x80..0xFF #Lo [11172] HANGUL SYLLABLE GA..HA...
|
|
| 0xEA 0xB1..0xFF 0x00..0xFF #
|
|
| 0xEB..0xEC 0x00..0xFF 0x00..0xFF #
|
|
| 0xED 0x00 0x00..0xFF #
|
|
| 0xED 0x01..0x9D 0x00..0xFF #
|
|
| 0xED 0x9E 0x00..0xA3 #
|
|
| 0xED 0x9E 0xB0..0xFF #Lo [23] HANGUL JUNGSEONG O-YEO..HANGUL JUN...
|
|
| 0xED 0x9F 0x00..0x86 #
|
|
| 0xED 0x9F 0x8B..0xBB #Lo [49] HANGUL JONGSEONG NIEUN-RIEUL..HANG...
|
|
| 0xEF 0xBE 0xA0..0xBE #Lo [31] HALFWIDTH HANGUL FILLER..HALFWIDTH...
|
|
| 0xEF 0xBF 0x82..0x87 #Lo [6] HALFWIDTH HANGUL LETTER A..HALFWID...
|
|
| 0xEF 0xBF 0x8A..0x8F #Lo [6] HALFWIDTH HANGUL LETTER YEO..HALFW...
|
|
| 0xEF 0xBF 0x92..0x97 #Lo [6] HALFWIDTH HANGUL LETTER YO..HALFWI...
|
|
| 0xEF 0xBF 0x9A..0x9C #Lo [3] HALFWIDTH HANGUL LETTER EU..HALFWI...
|
|
;
|
|
|
|
Han =
|
|
0xE2 0xBA 0x80..0x99 #So [26] CJK RADICAL REPEAT..CJK RADICAL RAP
|
|
| 0xE2 0xBA 0x9B..0xFF #So [89] CJK RADICAL CHOKE..CJK RADICAL C-S...
|
|
| 0xE2 0xBB 0x00..0xB3 #
|
|
| 0xE2 0xBC 0x80..0xFF #So [214] KANGXI RADICAL ONE..KANGXI RAD...
|
|
| 0xE2 0xBD..0xBE 0x00..0xFF #
|
|
| 0xE2 0xBF 0x00..0x95 #
|
|
| 0xE3 0x80 0x85 #Lm IDEOGRAPHIC ITERATION MARK
|
|
| 0xE3 0x80 0x87 #Nl IDEOGRAPHIC NUMBER ZERO
|
|
| 0xE3 0x80 0xA1..0xA9 #Nl [9] HANGZHOU NUMERAL ONE..HANGZHOU NUM...
|
|
| 0xE3 0x80 0xB8..0xBA #Nl [3] HANGZHOU NUMERAL TEN..HANGZHOU NUM...
|
|
| 0xE3 0x80 0xBB #Lm VERTICAL IDEOGRAPHIC ITERATION MARK
|
|
| 0xE3 0x90 0x80..0xFF #Lo [6582] CJK UNIFIED IDEOGRAPH-3400..C...
|
|
| 0xE3 0x91..0xFF 0x00..0xFF #
|
|
| 0xE4 0x00 0x00..0xFF #
|
|
| 0xE4 0x01..0xB5 0x00..0xFF #
|
|
| 0xE4 0xB6 0x00..0xB5 #
|
|
| 0xE4 0xB8 0x80..0xFF #Lo [20950] CJK UNIFIED IDEOGRAPH-...
|
|
| 0xE4 0xB9..0xFF 0x00..0xFF #
|
|
| 0xE5..0xE8 0x00..0xFF 0x00..0xFF #
|
|
| 0xE9 0x00 0x00..0xFF #
|
|
| 0xE9 0x01..0xBE 0x00..0xFF #
|
|
| 0xE9 0xBF 0x00..0x95 #
|
|
| 0xEF 0xA4 0x80..0xFF #Lo [366] CJK COMPATIBILITY IDEOGRAPH-F9...
|
|
| 0xEF 0xA5..0xA8 0x00..0xFF #
|
|
| 0xEF 0xA9 0x00..0xAD #
|
|
| 0xEF 0xA9 0xB0..0xFF #Lo [106] CJK COMPATIBILITY IDEOGRAPH-FA...
|
|
| 0xEF 0xAA..0xAA 0x00..0xFF #
|
|
| 0xEF 0xAB 0x00..0x99 #
|
|
| 0xF0 0xA0 0x80 0x80..0xFF #Lo [42711] CJK UNIFIED IDEOG...
|
|
| 0xF0 0xA0 0x81..0xFF 0x00..0xFF #
|
|
| 0xF0 0xA1..0xA9 0x00..0xFF 0x00..0xFF #
|
|
| 0xF0 0xAA 0x00 0x00..0xFF #
|
|
| 0xF0 0xAA 0x01..0x9A 0x00..0xFF #
|
|
| 0xF0 0xAA 0x9B 0x00..0x96 #
|
|
| 0xF0 0xAA 0x9C 0x80..0xFF #Lo [4149] CJK UNIFIED IDEOGRAPH-2A...
|
|
| 0xF0 0xAA 0x9D..0xFF 0x00..0xFF #
|
|
| 0xF0 0xAB 0x00 0x00..0xFF #
|
|
| 0xF0 0xAB 0x01..0x9B 0x00..0xFF #
|
|
| 0xF0 0xAB 0x9C 0x00..0xB4 #
|
|
| 0xF0 0xAB 0x9D 0x80..0xFF #Lo [222] CJK UNIFIED IDEOGRAPH-2B7...
|
|
| 0xF0 0xAB 0x9E..0x9F 0x00..0xFF #
|
|
| 0xF0 0xAB 0xA0 0x00..0x9D #
|
|
| 0xF0 0xAB 0xA0 0xA0..0xFF #Lo [5762] CJK UNIFIED IDEOGRAPH-2B...
|
|
| 0xF0 0xAB 0xA1..0xFF 0x00..0xFF #
|
|
| 0xF0 0xAC 0x00 0x00..0xFF #
|
|
| 0xF0 0xAC 0x01..0xB9 0x00..0xFF #
|
|
| 0xF0 0xAC 0xBA 0x00..0xA1 #
|
|
| 0xF0 0xAF 0xA0 0x80..0xFF #Lo [542] CJK COMPATIBILITY IDEOGRA...
|
|
| 0xF0 0xAF 0xA1..0xA7 0x00..0xFF #
|
|
| 0xF0 0xAF 0xA8 0x00..0x9D #
|
|
;
|
|
|
|
Hiragana =
|
|
0xE3 0x81 0x81..0xFF #Lo [86] HIRAGANA LETTER SMALL A..HIRAGANA ...
|
|
| 0xE3 0x82 0x00..0x96 #
|
|
| 0xE3 0x82 0x9D..0x9E #Lm [2] HIRAGANA ITERATION MARK..HIRAGANA ...
|
|
| 0xE3 0x82 0x9F #Lo HIRAGANA DIGRAPH YORI
|
|
| 0xF0 0x9B 0x80 0x81 #Lo HIRAGANA LETTER ARCHAIC YE
|
|
| 0xF0 0x9F 0x88 0x80 #So SQUARE HIRAGANA HOKA
|
|
;
|
|
|
|
}%%
|