X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/56786a220fd951f5e6dabb21ef70be1dd3b80389..d4442602046916e548aef5bd877fe0ec15b7c63d:/charset-unicode.inc.pl
diff --git a/charset-unicode.inc.pl b/charset-unicode.inc.pl
index 2b709c2..a0717a7 100644
--- a/charset-unicode.inc.pl
+++ b/charset-unicode.inc.pl
@@ -1,11 +1,12 @@
use utf8;
my %uniblock = (
- 0x000, '| control',
- 0x002, ' | comn',
- 0x004, ' | basic latin',
- 0x008, ' | control',
- 0x00A, ' | comn',
- 0x00C, ' | latin1',
+ # bmp
+ 0x000, ' | control',
+ 0x002, ' | comn',
+ 0x004, ' | basic latin',
+ 0x008, ' | control',
+ 0x00A, ' | comn',
+ 0x00C, ' | latin1',
0x010, ' | latin extended-A',
0x018, ' | latin extended-B',
0x020, ' | latin ext-B',
@@ -24,7 +25,8 @@ my %uniblock = (
0x07C, ' | n\'ko',
0x080, ' | samaritan',
0x084, ' | manda',
- 0x086, ' | reserved',
+ 0x086, ' | syr',
+ 0x087, ' | reserved',
0x08A, ' | arabic ext-A',
0x090, ' | devanagari',
0x098, ' | bengali',
@@ -69,8 +71,8 @@ my %uniblock = (
0x1BC, ' | batak',
0x1C0, ' | lepcha',
0x1C5, ' | ol chiki',
- 0x1C8, ' | cyr',
- 0x1C9, ' | reserved',
+ 0x1C8, ' | cyr',
+ 0x1C9, ' | georg+',
0x1CC, ' | sn',
0x1CD, ' | vedic',
0x1D0, ' | phonetic',
@@ -126,15 +128,15 @@ my %uniblock = (
0x320, ' | enclosed cjk characters',
0x330, ' | cjk compatibility',
0x340, ' | cjk unified ideographs extension A',
- 0x4D0, ' | cjk unified ideographs extension A',
+ 0x4D0, ' | cjk unified ideographs extension A',
0x4DC, ' | hexagrams',
0x4E0, ' | cjk unified ideographs',
0xA00, ' | yi',
- 0xA40, ' | yi',
+ 0xA40, ' | yi',
0xA49, ' | yi radicals',
0xA4D, ' | lisu',
0xA50, ' | vai',
- 0xA60, ' | vai',
+ 0xA60, ' | vai',
0xA64, ' | cyrillic extended-B',
0xA6A, ' | bamum',
0xA70, ' | tones',
@@ -155,21 +157,21 @@ my %uniblock = (
0xAAE, ' | mtei+',
0xAB0, ' | ethiopic-A',
0xAB3, ' | latin ext-E',
- 0xAB7, ' | cherokee+',
+ 0xAB7, ' | cherokee+',
0xABC, ' | meithei',
0xAC0, ' | hangeul syllables',
- 0xD70, ' | hangeul syllables',
+ 0xD70, ' | hangeul syllables',
0xD7B, ' | haungeul jamo-B',
0xD80, ' | high surrogates',
0xDC0, ' | low surrogates',
0xE00, ' | private use',
0xF90, ' | cjk compatibility ideographs',
0xFB0, ' | presentation',
- 0xFB5, ' | ',
- 0xFC0, ' | arabic presentation forms A',
- 0xFD0, ' | ',
+ 0xFB5, ' | ',
+ 0xFC0, ' | arabic presentation forms A',
+ 0xFD0, ' | ',
0xFDD, ' | ?',
- 0xFDF, ' | ',
+ 0xFDF, ' | ',
0xFE0, ' | var',
0xFE1, ' | ver',
0xFE2, ' | ½',
@@ -178,6 +180,262 @@ my %uniblock = (
0xFE7, ' | arabic presentation B',
0xFF0, ' | halfwidth & fullwidth forms',
0xFFF, ' | sp',
+
+ # smp
+ 0x1000, ' | linear B syllabary',
+ 0x1008, ' | linear B ideograms',
+ 0x1010, ' | aegean num',
+ 0x1014, ' | greek numbers',
+ 0x1019, ' | ancient sym',
+ 0x101D, ' | phaistos',
+ 0x1020, ' | iberian',
+ 0x1024, ' | reserved',
+ 0x1028, ' | lycian',
+ 0x102A, ' | carian',
+ 0x102E, ' | coptic',
+ 0x1030, ' | italic',
+ 0x1033, ' | gothic',
+ 0x1035, ' | permic',
+ 0x1038, ' | ugarit',
+ 0x103A, ' | old persian',
+ 0x103E, ' | sh.qs',
+ 0x1040, ' | deseret',
+ 0x1045, ' | shavian',
+ 0x1048, ' | osmanya',
+ 0x104B, ' | osage',
+ 0x1050, ' | elbasan',
+ 0x1053, ' | c albanian',
+ 0x1057, ' | vithkuqi',
+ 0x105C, ' | todhri',
+ 0x1060, ' | linear A',
+ 0x1070, ' | linear A',
+ 0x1078, ' | cypro-minoan',
+ 0x1080, ' | cypriot',
+ 0x1084, ' | aram',
+ 0x1086, ' | palmr',
+ 0x1088, ' | nabataean',
+ 0x108A, ' | res',
+ 0x108C, ' | numid',
+ 0x108E, ' | hatr',
+ 0x1090, ' | phoen',
+ 0x1092, ' | lydian',
+ 0x1094, ' | reserved',
+ 0x1098, ' | mero h',
+ 0x109A, ' | meroitic cursive',
+ 0x10A0, ' | kharoshthi',
+ 0x10A6, ' | s arab',
+ 0x10A8, ' | n arab',
+ 0x10AA, ' | balti',
+ 0x10AC, ' | manichaean',
+ 0x10B0, ' | avestan',
+ 0x10B4, ' | parth',
+ 0x10B6, ' | pahlav',
+ 0x10B8, ' | psalt pahl',
+ 0x10BB, ' | book pahl',
+ 0x10BE, ' | babur',
+ 0x10C0, ' | old turkic',
+ 0x10C5, ' | reserved',
+ 0x10C8, ' | old hungarian',
+ 0x10D0, ' | rohingya',
+ 0x10D4, ' | garay',
+ 0x10D8, ' | byblos',
+ 0x10E0, ' | reserved',
+ 0x10E6, ' | rumi',
+ 0x10E8, ' | reserved',
+ 0x10EE, ' | elym',
+ 0x10F0, ' | old sog',
+ 0x10F3, ' | sogdian',
+ 0x10F7, ' | res',
+ 0x10F8, ' | uyghur',
+ 0x1100, ' | brahmi',
+ 0x1108, ' | kaithi',
+ 0x110D, ' | sora som',
+ 0x1110, ' | chakma',
+ 0x1115, ' | mahajani',
+ 0x1118, ' | sharada',
+ 0x111E, ' | sinhal',
+ 0x1120, ' | khojki',
+ 0x1125, ' | landa',
+ 0x1128, ' | multani',
+ 0x112B, ' | khudabadi',
+ 0x1130, ' | grantha',
+ 0x1138, ' | tulu',
+ 0x113E, ' | shar+',
+ 0x1140, ' | newar',
+ 0x1148, ' | tirhuta',
+ 0x114E, ' | tani',
+ 0x1150, ' | ranjana',
+ 0x1158, ' | siddham',
+ 0x1160, ' | modi',
+ 0x1166, ' | mong',
+ 0x1168, ' | takri',
+ 0x116D, ' | jenticha',
+ 0x1170, ' | ahom',
+ 0x1174, ' | zou',
+ 0x117A, ' | pyu',
+ 0x1180, ' | dogra',
+ 0x1185, ' | sirmauri',
+ 0x1189, ' | res',
+ 0x118A, ' | warang citi',
+ 0x1190, ' | tolong siki',
+ 0x1194, ' | tikamuli',
+ 0x1198, ' | khambu rai',
+ 0x119C, ' | kirat rai',
+ 0x11A0, ' | zanabazar square',
+ 0x11A5, ' | soyombo',
+ 0x11AB, ' | res',
+ 0x11AC, ' | pau cin hau',
+ 0x11B0, ' | dhives akuru',
+ 0x11B5, ' | leke',
+ 0x11B9, ' | nandinagari',
+ 0x11BF, ' | res',
+ 0x11C0, ' | bhaiksuki',
+ 0x11C7, ' | marchen',
+ 0x11CC, ' | balti B',
+ 0x11D0, ' | masaram gondi',
+ 0x11D6, ' | gunjala gondi',
+ 0x11DB, ' | kawi',
+ 0x11E0, ' | tocharian',
+ 0x11E7, ' | khotanese',
+ 0x11ED, ' | res',
+ 0x11EE, ' | makas',
+ 0x11F0, ' | vatteluttu',
+ 0x11F4, ' | res',
+ 0x11F6, ' | chola',
+ 0x11FC, ' | tamil+',
+ 0x1200, ' | cuneiform',
+ 0x1240, ' | cuneiform numbers',
+ 0x1248, ' | early dynastic cuneiform',
+ 0x1250, ' | e. dyn. cuneiform',
+ 0x1255, ' | reserved',
+ 0x1260, ' | proto-cuneiform',
+ 0x12E0, ' | indus',
+ 0x12F0, ' | indus',
+ 0x12F9, ' | reserved',
+ 0x1300, ' | egyptian hieroglyphs',
+ 0x1340, ' | egyptian',
+ 0x1343, ' | eg.c',
+ 0x1344, ' | ',
+ 0x1350, ' | egyptian hieroglyphs extended-A',
+ 0x1440, ' | anatolian hieroglyphs',
+ 0x1460, ' | anatolian',
+ 0x1468, ' | ',
+ 0x1470, ' | egyptian hieroglyphs extended-B',
+ 0x1500, ' | lampung',
+ 0x1504, ' | kerinci',
+ 0x1507, ' | res',
+ 0x1508, ' | ',
+ 0x1510, ' | mandombe',
+ 0x1550, ' | maya hieroglyphs',
+ 0x15A0, ' | reserved',
+ 0x15C0, ' | aztec pictograms',
+ 0x1600, ' | cirth',
+ 0x1608, ' | tengwar',
+ 0x1610, ' | khema',
+ 0x1614, ' | khe prih',
+ 0x1618, ' | res',
+ 0x161A, ' | moon',
+ 0x1620, ' | blissymbols',
+ 0x1670, ' | bagam',
+ 0x167B, ' | iban',
+ 0x1680, ' | bamum supplement',
+ 0x16A0, ' | bamum+',
+ 0x16A4, ' | mro',
+ 0x16A7, ' | mossang tangsa',
+ 0x16AD, ' | bassa vah',
+ 0x16B0, ' | pahawh hmong',
+ 0x16B9, ' | woleai',
+ 0x16C0, ' | kpelle',
+ 0x16C8, ' | afaka',
+ 0x16CD, ' | lk tangsa',
+ 0x16D0, ' | reserved',
+ 0x16DD, ' | kulitan',
+ 0x16E0, ' | mwangwego',
+ 0x16E4, ' | medefaidrin',
+ 0x16EA, ' | lontara+',
+ 0x16F0, ' | miao',
+ 0x16FA, ' | lontara b-b',
+ 0x16FE, ' | ideo',
+ 0x1700, ' | tangut ideographs',
+ 0x1880, ' | tangut components',
+ 0x18B0, ' | khitan small',
+ 0x18D0, ' | khitan ideographs',
+ 0x1960, ' | jurchen',
+ 0x19B0, ' | jurchen',
+ 0x19B6, ' | jurchen rad',
+ 0x19BA, ' | reserved',
+ 0x19C0, ' | reserved',
+ 0x19E0, ' | pau cin hau syllabary',
+ 0x1A30, ' | eskaya',
+ 0x1A80, ' | naxi geba',
+ 0x1AA0, ' | naxi geba',
+ 0x1AAB, ' | res',
+ 0x1AAC, ' | ',
+ 0x1AB0, ' | naxi dongba',
+ 0x1B00, ' | kana supplement',
+ 0x1B10, ' | kana+A',
+ 0x1B13, ' | kaidÄ',
+ 0x1B17, ' | ',
+ 0x1B20, ' | nushu',
+ 0x1B30, ' | shuishu',
+ 0x1B50, ' | proto-elamite',
+ 0x1BC0, ' | duployan',
+ 0x1BCA, ' | sh',
+ 0x1BCB, ' | pitman',
+ 0x1BD0, ' | shorthands?',
+ 0x1C00, ' | micmac hieroglyphs',
+ 0x1CB0, ' | rongorongo',
+ 0x1CE0, ' | reserved',
+ 0x1D00, ' | byzantine musical',
+ 0x1D10, ' | musical symbols',
+ 0x1D20, ' | anc greek music',
+ 0x1D25, ' | reserved',
+ 0x1D2E, ' | mayan',
+ 0x1D30, ' | tai xuan jing',
+ 0x1D36, ' | rod',
+ 0x1D38, ' | mathematical alphanumeric+',
+ 0x1D40, ' | mathematical alphanumeric',
+ 0x1D80, ' | sutton',
+ 0x1DA0, ' | ',
+ 0x1DAB, ' | reserved',
+ 0x1DB0, ' | reserved',
+ 0x1E00, ' | glagol+',
+ 0x1E03, ' | pallava',
+ 0x1E08, ' | chalukya',
+ 0x1E0E, ' | res',
+ 0x1E10, ' | eebee hmong',
+ 0x1E1B, ' | cher vang hmong',
+ 0x1E20, ' | western cham',
+ 0x1E27, ' | beria',
+ 0x1E2A, ' | reserved',
+ 0x1E30, ' | loma',
+ 0x1E50, ' | reserved',
+ 0x1E80, ' | mende kikakui',
+ 0x1E8E, ' | res',
+ 0x1E90, ' | adlam',
+ 0x1E96, ' | reserved',
+ 0x1EA0, ' | reserved',
+ 0x1EC0, ' | persian siyaq',
+ 0x1EC7, ' | indic siyaq',
+ 0x1ECC, ' | diwani siyaq',
+ 0x1ED0, ' | ottoman siyaq',
+ 0x1ED5, ' | reserved',
+ 0x1EE0, ' | arabic mathematical alphabetic', # Sm
+ 0x1EF0, ' | reserved',
+ 0x1F00, ' | mahjong',
+ 0x1F03, ' | domino tiles',
+ 0x1F0A, ' | playing cards',
+ 0x1F10, ' | enclosed alphanumeric supplement', # So
+ 0x1F20, ' | enclosed ideographic supplement', # So
+ 0x1F30, ' | miscellaneous symbols and pictographs',
+ 0x1F60, ' | emoticons',
+ 0x1F65, ' | ornament',
+ 0x1F68, ' | transport',
+ 0x1F70, ' | alchemical',
+ 0x1F78, ' | geometric shapes ext',
+ 0x1F80, ' | supplemental arrows-C',
+ 0x1F90, ' | supplemental symbols and pictographs',
+ 0x1FA0, ' | reserved',
);
sub {
|