d1608b447d
It makes compatible with IANA charset defination, and let existent I18N app happier. Ref: <http://www.isi.edu/in-notes/iana/assignments/character-sets>
196 lines
7.0 KiB
Plaintext
196 lines
7.0 KiB
Plaintext
/*
|
|
* big5 first byte: A1-FE
|
|
* second byte: 40-7E, A1-FE
|
|
*/
|
|
|
|
ENCODING "BIG5"
|
|
/* VARIABLE BIG5 character set */
|
|
|
|
/*
|
|
* ASCII
|
|
*/
|
|
ALPHA 'A' - 'Z' 'a' - 'z'
|
|
CONTROL 0x00 - 0x1f 0x7f
|
|
DIGIT '0' - '9'
|
|
GRAPH 0x21 - 0x7e
|
|
LOWER 'a' - 'z'
|
|
PUNCT 0x21 - 0x2f 0x3a - 0x40 0x5b - 0x60 0x7b - 0x7e
|
|
SPACE 0x09 - 0x0d 0x20
|
|
UPPER 'A' - 'Z'
|
|
XDIGIT '0' - '9' 'a' - 'f' 'A' - 'F'
|
|
BLANK ' ' '\t'
|
|
PRINT 0x20 - 0x7e
|
|
|
|
MAPLOWER < 'A' - 'Z' : 'a' >
|
|
MAPLOWER < 'a' - 'z' : 'a' >
|
|
MAPUPPER < 'A' - 'Z' : 'A' >
|
|
MAPUPPER < 'a' - 'z' : 'A' >
|
|
TODIGIT < '0' - '9' : 0 >
|
|
TODIGIT < 'A' - 'F' : 10 >
|
|
TODIGIT < 'a' - 'f' : 10 >
|
|
|
|
/*
|
|
* the real thing
|
|
*/
|
|
|
|
PUNCT 0xa141 - 0xa17e 0xa1a1 - 0xa1ac
|
|
|
|
SPECIAL 0xa1ad - 0xa1fe 0xa240 - 0xa27e 0xa2a1 - 0xa2ae
|
|
|
|
/* full width 0 1 2 .. 9 */
|
|
DIGIT 0xa2af - 0xa2b8
|
|
|
|
/* map these (SuChou Code (12) and Roman (I II etc.) number (10)) to digits */
|
|
TODIGIT < 0xa2b9 - 0xa2c2 : 1 >
|
|
TODIGIT < 0xa2c3 - 0xa2ce : 1 >
|
|
|
|
/* full width A .. Z and a .. z */
|
|
/* note the lower case alphabets are not continuous */
|
|
UPPER 0xa2cf - 0xa2e8
|
|
LOWER 0xa2e9 - 0xa2fe 0xa340 - 0xa343
|
|
|
|
/* we have to got two line for each 'cause they are not continuous */
|
|
MAPLOWER < 0xa2cf - 0xa2e4 : 0xa2e9 > < 0xa2e5 - 0xa2e8 : 0xa340 >
|
|
MAPLOWER < 0xa2e9 - 0xa2fe : 0xa2e9 > < 0xa340 - 0xa343 : 0xa340 >
|
|
MAPUPPER < 0xa2cf - 0xa2e4 : 0xa2cf > < 0xa2e5 - 0xa2e8 : 0xa2e5 >
|
|
MAPUPPER < 0xa2e9 - 0xa2fe : 0xa2cf > < 0xa340 - 0xa343 : 0xa2e5 >
|
|
|
|
XDIGIT 0xa2cf - 0xa2d4 0xa2e9 - 0xa2ee
|
|
|
|
SPACE 0xa3bc
|
|
/* bopomofo symbols */
|
|
PHONOGRAM 0xa374 - 0xa37e 0xa3a1 - 0xa3bb
|
|
/* tone symbols */
|
|
PHONOGRAM 0xa3bd - 0xa3bf
|
|
|
|
/* greek */
|
|
UPPER 0xa344 - 0xa35b
|
|
LOWER 0xa35c - 0xa373
|
|
MAPUPPER < 0xa344 - 0xa35b : 0xa344 > < 0xa35c - 0xa373 : 0xa344 >
|
|
MAPLOWER < 0xa344 - 0xa35b : 0xa35c > < 0xa35c - 0xa373 : 0xa35c >
|
|
|
|
/* cyrillic (russian etc) */
|
|
UPPER 0xc7f3 - 0xc7fe 0xc840 - 0xc854
|
|
LOWER 0xc855 - 0xc875
|
|
MAPUPPER < 0xc7f3 - 0xc7fe : 0xc7f3 > < 0xc840 - 0xc854 : 0xc840 >
|
|
MAPUPPER < 0xc855 - 0xc860 : 0xc7f3 > < 0xc861 - 0xc875 : 0xc840 >
|
|
MAPLOWER < 0xc7f3 - 0xc7fe : 0xc855 > < 0xc840 - 0xc854 : 0xc861 >
|
|
MAPLOWER < 0xc855 - 0xc860 : 0xc855 > < 0xc861 - 0xc875 : 0xc861 >
|
|
|
|
IDEOGRAM 0xa440 - 0xa47e 0xa4a1 - 0xa4fe
|
|
IDEOGRAM 0xa540 - 0xa57e 0xa5a1 - 0xa5fe
|
|
IDEOGRAM 0xa640 - 0xa67e 0xa6a1 - 0xa6fe
|
|
IDEOGRAM 0xa740 - 0xa77e 0xa7a1 - 0xa7fe
|
|
IDEOGRAM 0xa840 - 0xa87e 0xa8a1 - 0xa8fe
|
|
IDEOGRAM 0xa940 - 0xa97e 0xa9a1 - 0xa9fe
|
|
IDEOGRAM 0xaa40 - 0xaa7e 0xaaa1 - 0xaafe
|
|
IDEOGRAM 0xab40 - 0xab7e 0xaba1 - 0xabfe
|
|
IDEOGRAM 0xac40 - 0xac7e 0xaca1 - 0xacfe
|
|
IDEOGRAM 0xad40 - 0xad7e 0xada1 - 0xadfe
|
|
IDEOGRAM 0xae40 - 0xae7e 0xaea1 - 0xaefe
|
|
IDEOGRAM 0xaf40 - 0xaf7e 0xafa1 - 0xaffe
|
|
IDEOGRAM 0xb040 - 0xb07e 0xb0a1 - 0xb0fe
|
|
IDEOGRAM 0xb140 - 0xb17e 0xb1a1 - 0xb1fe
|
|
IDEOGRAM 0xb240 - 0xb27e 0xb2a1 - 0xb2fe
|
|
IDEOGRAM 0xb340 - 0xb37e 0xb3a1 - 0xb3fe
|
|
IDEOGRAM 0xb440 - 0xb47e 0xb4a1 - 0xb4fe
|
|
IDEOGRAM 0xb540 - 0xb57e 0xb5a1 - 0xb5fe
|
|
IDEOGRAM 0xb640 - 0xb67e 0xb6a1 - 0xb6fe
|
|
IDEOGRAM 0xb740 - 0xb77e 0xb7a1 - 0xb7fe
|
|
IDEOGRAM 0xb840 - 0xb87e 0xb8a1 - 0xb8fe
|
|
IDEOGRAM 0xb940 - 0xb97e 0xb9a1 - 0xb9fe
|
|
IDEOGRAM 0xba40 - 0xba7e 0xbaa1 - 0xbafe
|
|
IDEOGRAM 0xbb40 - 0xbb7e 0xbba1 - 0xbbfe
|
|
IDEOGRAM 0xbc40 - 0xbc7e 0xbca1 - 0xbcfe
|
|
IDEOGRAM 0xbd40 - 0xbd7e 0xbda1 - 0xbdfe
|
|
IDEOGRAM 0xbe40 - 0xbe7e 0xbea1 - 0xbefe
|
|
IDEOGRAM 0xbf40 - 0xbf7e 0xbfa1 - 0xbffe
|
|
IDEOGRAM 0xc040 - 0xc07e 0xc0a1 - 0xc0fe
|
|
IDEOGRAM 0xc140 - 0xc17e 0xc1a1 - 0xc1fe
|
|
IDEOGRAM 0xc240 - 0xc27e 0xc2a1 - 0xc2fe
|
|
IDEOGRAM 0xc340 - 0xc37e 0xc3a1 - 0xc3fe
|
|
IDEOGRAM 0xc440 - 0xc47e 0xc4a1 - 0xc4fe
|
|
IDEOGRAM 0xc540 - 0xc57e 0xc5a1 - 0xc5fe
|
|
IDEOGRAM 0xc640 - 0xc67e
|
|
|
|
IDEOGRAM 0xc940 - 0xc97e 0xc9a1 - 0xc9fe
|
|
IDEOGRAM 0xca40 - 0xca7e 0xcaa1 - 0xcafe
|
|
IDEOGRAM 0xcb40 - 0xcb7e 0xcba1 - 0xcbfe
|
|
IDEOGRAM 0xcc40 - 0xcc7e 0xcca1 - 0xccfe
|
|
IDEOGRAM 0xcd40 - 0xcd7e 0xcda1 - 0xcdfe
|
|
IDEOGRAM 0xce40 - 0xce7e 0xcea1 - 0xcefe
|
|
IDEOGRAM 0xcf40 - 0xcf7e 0xcfa1 - 0xcffe
|
|
IDEOGRAM 0xd040 - 0xd074 0xd0a1 - 0xd0fe
|
|
IDEOGRAM 0xd140 - 0xd174 0xd1a1 - 0xd1fe
|
|
IDEOGRAM 0xd240 - 0xd274 0xd2a1 - 0xd2fe
|
|
IDEOGRAM 0xd340 - 0xd374 0xd3a1 - 0xd3fe
|
|
IDEOGRAM 0xd440 - 0xd474 0xd4a1 - 0xd4fe
|
|
IDEOGRAM 0xd540 - 0xd574 0xd5a1 - 0xd5fe
|
|
IDEOGRAM 0xd640 - 0xd674 0xd6a1 - 0xd6fe
|
|
IDEOGRAM 0xd740 - 0xd774 0xd7a1 - 0xd7fe
|
|
IDEOGRAM 0xd840 - 0xd874 0xd8a1 - 0xd8fe
|
|
IDEOGRAM 0xd940 - 0xd974 0xd9a1 - 0xd9fe
|
|
IDEOGRAM 0xda40 - 0xda74 0xdaa1 - 0xdafe
|
|
IDEOGRAM 0xdb40 - 0xdb74 0xdba1 - 0xdbfe
|
|
IDEOGRAM 0xdc40 - 0xdc74 0xdca1 - 0xdcfe
|
|
IDEOGRAM 0xdd40 - 0xdd74 0xdda1 - 0xddfe
|
|
IDEOGRAM 0xde40 - 0xde74 0xdea1 - 0xdefe
|
|
IDEOGRAM 0xdf40 - 0xdf74 0xdfa1 - 0xdffe
|
|
IDEOGRAM 0xe040 - 0xe074 0xe0a1 - 0xe0fe
|
|
IDEOGRAM 0xe140 - 0xe174 0xe1a1 - 0xe1fe
|
|
IDEOGRAM 0xe240 - 0xe274 0xe2a1 - 0xe2fe
|
|
IDEOGRAM 0xe340 - 0xe374 0xe3a1 - 0xe3fe
|
|
IDEOGRAM 0xe440 - 0xe474 0xe4a1 - 0xe4fe
|
|
IDEOGRAM 0xe540 - 0xe574 0xe5a1 - 0xe5fe
|
|
IDEOGRAM 0xe640 - 0xe674 0xe6a1 - 0xe6fe
|
|
IDEOGRAM 0xe740 - 0xe774 0xe7a1 - 0xe7fe
|
|
IDEOGRAM 0xe840 - 0xe874 0xe8a1 - 0xe8fe
|
|
IDEOGRAM 0xe940 - 0xe974 0xe9a1 - 0xe9fe
|
|
IDEOGRAM 0xea40 - 0xea74 0xeaa1 - 0xeafe
|
|
IDEOGRAM 0xeb40 - 0xeb74 0xeba1 - 0xebfe
|
|
IDEOGRAM 0xec40 - 0xec74 0xeca1 - 0xecfe
|
|
IDEOGRAM 0xed40 - 0xed74 0xeda1 - 0xedfe
|
|
IDEOGRAM 0xee40 - 0xee74 0xeea1 - 0xeefe
|
|
IDEOGRAM 0xef40 - 0xef74 0xefa1 - 0xeffe
|
|
IDEOGRAM 0xf040 - 0xf074 0xf0a1 - 0xf0fe
|
|
IDEOGRAM 0xf140 - 0xf174 0xf1a1 - 0xf1fe
|
|
IDEOGRAM 0xf240 - 0xf274 0xf2a1 - 0xf2fe
|
|
IDEOGRAM 0xf340 - 0xf374 0xf3a1 - 0xf3fe
|
|
IDEOGRAM 0xf440 - 0xf474 0xf4a1 - 0xf4fe
|
|
IDEOGRAM 0xf540 - 0xf574 0xf5a1 - 0xf5fe
|
|
IDEOGRAM 0xf640 - 0xf674 0xf6a1 - 0xf6fe
|
|
IDEOGRAM 0xf740 - 0xf774 0xf7a1 - 0xf7fe
|
|
IDEOGRAM 0xf840 - 0xf874 0xf8a1 - 0xf8fe
|
|
IDEOGRAM 0xf940 - 0xf974 0xf9a1 - 0xf9dc
|
|
|
|
/* Eten external charset 0xc6a1 - 0xc8d3 */
|
|
|
|
/* lots strange numbers */
|
|
SPECIAL 0xc6a1 - 0xc6be
|
|
|
|
TODIGIT < 0xc6a1 - 0xc6aa : 1 > < 0xc6ab - 0xc6b4 : 1 >
|
|
TODIGIT < 0xc6b5 - 0xc6be : 1>
|
|
|
|
/* these are HanZi Radicals */
|
|
SPECIAL 0xc6bf - 0xc6d7 0xc879 - 0xc87e 0xc8a1 - 0xc8a4
|
|
|
|
/* who knows what these are ... */
|
|
SPECIAL 0xc6d8 - 0xc6e6 0xc876 - 0xc878
|
|
|
|
/* hiragana */
|
|
SPECIAL 0xc6e7 - 0xc6fe 0xc740 - 0xc77a
|
|
|
|
/* katakana */
|
|
SPECIAL 0xc77b - 0xc77e 0xc7a1 - 0xc7f2
|
|
|
|
/* ZIP code (in Japan) */
|
|
SPECIAL 0xc8a5 - 0x08cc
|
|
|
|
/* various Japanese symbols */
|
|
SPECIAL 0x08cd - 0xc8d3
|
|
|
|
/* line drawing glyphs */
|
|
SPECIAL 0xf9dd - 0xf9ff
|
|
|
|
|