summaryrefslogtreecommitdiff
path: root/Build/source/texk/web2c/uptexdir/kanji.c
diff options
context:
space:
mode:
authorTakuji Tanaka <KXD02663@nifty.ne.jp>2014-12-30 00:48:13 +0000
committerTakuji Tanaka <KXD02663@nifty.ne.jp>2014-12-30 00:48:13 +0000
commitbf6f668244ad61c477919b5a1f511960c6178efd (patch)
treed7f9ce1b534dff1aca8f8d56ea69da5c898d35e5 /Build/source/texk/web2c/uptexdir/kanji.c
parent2dfc237830e825534c39e178688ca627c561db68 (diff)
upTeX 1.20
git-svn-id: svn://tug.org/texlive/trunk@35923 c570f23f-e606-0410-a88d-b1316a301751
Diffstat (limited to 'Build/source/texk/web2c/uptexdir/kanji.c')
-rw-r--r--Build/source/texk/web2c/uptexdir/kanji.c134
1 files changed, 83 insertions, 51 deletions
diff --git a/Build/source/texk/web2c/uptexdir/kanji.c b/Build/source/texk/web2c/uptexdir/kanji.c
index 86d80226832..a8ab7b5938d 100644
--- a/Build/source/texk/web2c/uptexdir/kanji.c
+++ b/Build/source/texk/web2c/uptexdir/kanji.c
@@ -60,8 +60,8 @@ integer calc_pos(integer c)
}
/* Ref. http://www.unicode.org/Public/UNIDATA/Blocks.txt */
-/* # Blocks-6.1.0.txt */
-/* # Date: 2011-06-14, 18:26:00 GMT [KW, LI] */
+/* # Blocks-7.0.0.txt */
+/* # Date: 2014-04-03, 23:23:00 GMT [RP, KW] */
static long ucs_range[]={
0x0000, /* Basic Latin 0x00 */
0x0080, /* Latin-1 Supplement */
@@ -118,168 +118,200 @@ static long ucs_range[]={
0x19E0, /* Khmer Symbols */
0x1A00, /* Buginese */
0x1A20, /* Tai Tham */
- 0x1B00, /* Balinese */
- 0x1B80, /* Sundanese 0x38 */
+ 0x1AB0, /* Combining Diacritical Marks Extended */
+ 0x1B00, /* Balinese 0x38 */
+ 0x1B80, /* Sundanese */
0x1BC0, /* Batak */
0x1C00, /* Lepcha */
0x1C50, /* Ol Chiki */
0x1CC0, /* Sundanese Supplement */
0x1CD0, /* Vedic Extensions */
0x1D00, /* Phonetic Extensions */
- 0x1D80, /* Phonetic Extensions Supplement */
- 0x1DC0, /* Combining Diacritical Marks Supplement 0x40 */
+ 0x1D80, /* Phonetic Extensions Supplement 0x40 */
+ 0x1DC0, /* Combining Diacritical Marks Supplement */
0x1E00, /* Latin Extended Additional */
0x1F00, /* Greek Extended */
0x2000, /* General Punctuation */
0x2070, /* Superscripts and Subscripts */
0x20A0, /* Currency Symbols */
0x20D0, /* Combining Diacritical Marks for Symbols */
- 0x2100, /* Letterlike Symbols */
- 0x2150, /* Number Forms 0x48 */
+ 0x2100, /* Letterlike Symbols 0x48 */
+ 0x2150, /* Number Forms */
0x2190, /* Arrows */
0x2200, /* Mathematical Operators */
0x2300, /* Miscellaneous Technical */
0x2400, /* Control Pictures */
0x2440, /* Optical Character Recognition */
0x2460, /* Enclosed Alphanumerics */
- 0x2500, /* Box Drawing */
- 0x2580, /* Block Elements 0x50 */
+ 0x2500, /* Box Drawing 0x50 */
+ 0x2580, /* Block Elements */
0x25A0, /* Geometric Shapes */
0x2600, /* Miscellaneous Symbols */
0x2700, /* Dingbats */
0x27C0, /* Miscellaneous Mathematical Symbols-A */
0x27F0, /* Supplemental Arrows-A */
0x2800, /* Braille Patterns */
- 0x2900, /* Supplemental Arrows-B */
- 0x2980, /* Miscellaneous Mathematical Symbols-B 0x58 */
+ 0x2900, /* Supplemental Arrows-B 0x58 */
+ 0x2980, /* Miscellaneous Mathematical Symbols-B */
0x2A00, /* Supplemental Mathematical Operators */
0x2B00, /* Miscellaneous Symbols and Arrows */
0x2C00, /* Glagolitic */
0x2C60, /* Latin Extended-C */
0x2C80, /* Coptic */
0x2D00, /* Georgian Supplement */
- 0x2D30, /* Tifinagh */
- 0x2D80, /* Ethiopic Extended 0x60 */
+ 0x2D30, /* Tifinagh 0x60 */
+ 0x2D80, /* Ethiopic Extended */
0x2DE0, /* Cyrillic Extended-A */
0x2E00, /* Supplemental Punctuation */
0x2E80, /* CJK Radicals Supplement */
0x2F00, /* Kangxi Radicals */
0x2FF0, /* Ideographic Description Characters */
0x3000, /* CJK Symbols and Punctuation */
- 0x3040, /* Hiragana */
- 0x30A0, /* Katakana 0x68 */
+ 0x3040, /* Hiragana 0x68 */
+ 0x30A0, /* Katakana */
0x3100, /* Bopomofo */
0x3130, /* Hangul Compatibility Jamo */
0x3190, /* Kanbun */
0x31A0, /* Bopomofo Extended */
0x31C0, /* CJK Strokes */
0x31F0, /* Katakana Phonetic Extensions */
- 0x3200, /* Enclosed CJK Letters and Months */
- 0x3300, /* CJK Compatibility 0x70 */
+ 0x3200, /* Enclosed CJK Letters and Months 0x70 */
+ 0x3300, /* CJK Compatibility */
0x3400, /* CJK Unified Ideographs Extension A */
0x4DC0, /* Yijing Hexagram Symbols */
0x4E00, /* CJK Unified Ideographs */
0xA000, /* Yi Syllables */
0xA490, /* Yi Radicals */
0xA4D0, /* Lisu */
- 0xA500, /* Vai */
- 0xA640, /* Cyrillic Extended-B 0x78 */
+ 0xA500, /* Vai 0x78 */
+ 0xA640, /* Cyrillic Extended-B */
0xA6A0, /* Bamum */
0xA700, /* Modifier Tone Letters */
0xA720, /* Latin Extended-D */
0xA800, /* Syloti Nagri */
0xA830, /* Common Indic Number Forms */
0xA840, /* Phags-pa */
- 0xA880, /* Saurashtra */
- 0xA8E0, /* Devanagari Extended 0x80 */
+ 0xA880, /* Saurashtra 0x80 */
+ 0xA8E0, /* Devanagari Extended */
0xA900, /* Kayah Li */
0xA930, /* Rejang */
0xA960, /* Hangul Jamo Extended-A */
0xA980, /* Javanese */
+ 0xA9E0, /* Myanmar Extended-B */
0xAA00, /* Cham */
- 0xAA60, /* Myanmar Extended-A */
+ 0xAA60, /* Myanmar Extended-A 0x88 */
0xAA80, /* Tai Viet */
- 0xAAE0, /* Meetei Mayek Extensions 0x88 */
+ 0xAAE0, /* Meetei Mayek Extensions */
0xAB00, /* Ethiopic Extended-A */
+ 0xAB30, /* Latin Extended-E */
0xABC0, /* Meetei Mayek */
0xAC00, /* Hangul Syllables */
0xD7B0, /* Hangul Jamo Extended-B */
- 0xD800, /* High Surrogates */
+ 0xD800, /* High Surrogates 0x90 */
0xDB80, /* High Private Use Surrogates */
0xDC00, /* Low Surrogates */
- 0xE000, /* Private Use Area 0x90 */
+ 0xE000, /* Private Use Area */
0xF900, /* CJK Compatibility Ideographs */
0xFB00, /* Alphabetic Presentation Forms */
0xFB50, /* Arabic Presentation Forms-A */
0xFE00, /* Variation Selectors */
- 0xFE10, /* Vertical Forms */
+ 0xFE10, /* Vertical Forms 0x98 */
0xFE20, /* Combining Half Marks */
0xFE30, /* CJK Compatibility Forms */
- 0xFE50, /* Small Form Variants 0x98 */
+ 0xFE50, /* Small Form Variants */
0xFE70, /* Arabic Presentation Forms-B */
0xFF00, /* Halfwidth and Fullwidth Forms */
0xFFF0, /* Specials */
0x10000, /* Linear B Syllabary */
- 0x10080, /* Linear B Ideograms */
+ 0x10080, /* Linear B Ideograms 0xa0 */
0x10100, /* Aegean Numbers */
0x10140, /* Ancient Greek Numbers */
- 0x10190, /* Ancient Symbols 0xa0 */
+ 0x10190, /* Ancient Symbols */
0x101D0, /* Phaistos Disc */
0x10280, /* Lycian */
0x102A0, /* Carian */
- 0x10300, /* Old Italic */
+ 0x102E0, /* Coptic Epact Numbers */
+ 0x10300, /* Old Italic 0xa8 */
0x10330, /* Gothic */
+ 0x10350, /* Old Permic */
0x10380, /* Ugaritic */
0x103A0, /* Old Persian */
- 0x10400, /* Deseret 0xa8 */
+ 0x10400, /* Deseret */
0x10450, /* Shavian */
0x10480, /* Osmanya */
+ 0x10500, /* Elbasan 0xb0 */
+ 0x10530, /* Caucasian Albanian */
+ 0x10600, /* Linear A */
0x10800, /* Cypriot Syllabary */
0x10840, /* Imperial Aramaic */
+ 0x10860, /* Palmyrene */
+ 0x10880, /* Nabataean */
0x10900, /* Phoenician */
- 0x10920, /* Lydian */
+ 0x10920, /* Lydian 0xb8 */
0x10980, /* Meroitic Hieroglyphs */
- 0x109A0, /* Meroitic Cursive 0xb0 */
+ 0x109A0, /* Meroitic Cursive */
0x10A00, /* Kharoshthi */
0x10A60, /* Old South Arabian */
+ 0x10A80, /* Old North Arabian */
+ 0x10AC0, /* Manichaean */
0x10B00, /* Avestan */
- 0x10B40, /* Inscriptional Parthian */
+ 0x10B40, /* Inscriptional Parthian 0xc0 */
0x10B60, /* Inscriptional Pahlavi */
+ 0x10B80, /* Psalter Pahlavi */
0x10C00, /* Old Turkic */
0x10E60, /* Rumi Numeral Symbols */
- 0x11000, /* Brahmi 0xb8 */
+ 0x11000, /* Brahmi */
0x11080, /* Kaithi */
0x110D0, /* Sora Sompeng */
- 0x11100, /* Chakma */
+ 0x11100, /* Chakma 0xc8 */
+ 0x11150, /* Mahajani */
0x11180, /* Sharada */
+ 0x111E0, /* Sinhala Archaic Numbers */
+ 0x11200, /* Khojki */
+ 0x112B0, /* Khudawadi */
+ 0x11300, /* Grantha */
+ 0x11480, /* Tirhuta */
+ 0x11580, /* Siddham 0xd0 */
+ 0x11600, /* Modi */
0x11680, /* Takri */
+ 0x118A0, /* Warang Citi */
+ 0x11AC0, /* Pau Cin Hau */
0x12000, /* Cuneiform */
0x12400, /* Cuneiform Numbers and Punctuation */
- 0x13000, /* Egyptian Hieroglyphs 0xc0 */
- 0x16800, /* Bamum Supplement */
+ 0x13000, /* Egyptian Hieroglyphs */
+ 0x16800, /* Bamum Supplement 0xd8 */
+ 0x16A40, /* Mro */
+ 0x16AD0, /* Bassa Vah */
+ 0x16B00, /* Pahawh Hmong */
0x16F00, /* Miao */
0x1B000, /* Kana Supplement */
- 0x1D000, /* Byzantine Musical Symbols */
+ 0x1BC00, /* Duployan */
+ 0x1BCA0, /* Shorthand Format Controls */
+ 0x1D000, /* Byzantine Musical Symbols 0xe0 */
0x1D100, /* Musical Symbols */
0x1D200, /* Ancient Greek Musical Notation */
0x1D300, /* Tai Xuan Jing Symbols */
- 0x1D360, /* Counting Rod Numerals 0xc8 */
+ 0x1D360, /* Counting Rod Numerals */
0x1D400, /* Mathematical Alphanumeric Symbols */
+ 0x1E800, /* Mende Kikakui */
0x1EE00, /* Arabic Mathematical Alphabetic Symbols */
- 0x1F000, /* Mahjong Tiles */
+ 0x1F000, /* Mahjong Tiles 0xe8 */
0x1F030, /* Domino Tiles */
0x1F0A0, /* Playing Cards */
0x1F100, /* Enclosed Alphanumeric Supplement */
0x1F200, /* Enclosed Ideographic Supplement */
- 0x1F300, /* Miscellaneous Symbols And Pictographs 0xd0 */
+ 0x1F300, /* Miscellaneous Symbols and Pictographs */
0x1F600, /* Emoticons */
- 0x1F680, /* Transport And Map Symbols */
+ 0x1F650, /* Ornamental Dingbats */
+ 0x1F680, /* Transport and Map Symbols 0xf0 */
0x1F700, /* Alchemical Symbols */
+ 0x1F780, /* Geometric Shapes Extended */
+ 0x1F800, /* Supplemental Arrows-C */
0x20000, /* CJK Unified Ideographs Extension B */
0x2A700, /* CJK Unified Ideographs Extension C */
0x2B740, /* CJK Unified Ideographs Extension D */
0x2F800, /* CJK Compatibility Ideographs Supplement */
- 0x30000, /* reserved 0xd8 */
+ 0x30000, /* reserved 0xf8 */
0x40000, /* reserved */
0x50000, /* reserved */
0x60000, /* reserved */
@@ -287,7 +319,7 @@ static long ucs_range[]={
0x80000, /* reserved */
0x90000, /* reserved */
0xA0000, /* reserved */
- 0xB0000, /* reserved 0xe0 */
+ 0xB0000, /* reserved 0x100 */
0xC0000, /* reserved */
0xD0000, /* reserved */
0xE0000, /* Tags */
@@ -297,7 +329,7 @@ static long ucs_range[]={
/* Value over 0x10FFFF is illegal under Unicode,
They are for some special use. *** experimental *** */
0x110000, /* Reserved */
- 0x120000, /* Reserved 0xe8 */
+ 0x120000, /* Reserved 0x108 */
0x130000, /* Reserved */
0x140000, /* Reserved */
0x150000, /* Reserved */
@@ -305,7 +337,7 @@ static long ucs_range[]={
0x170000, /* Reserved */
0x180000, /* Reserved */
0x190000, /* Reserved */
- 0x1A0000, /* Reserved 0xf0 */
+ 0x1A0000, /* Reserved 0x110 */
0x1B0000, /* Reserved */
0x1C0000, /* Reserved */
0x1D0000, /* Reserved */
@@ -313,7 +345,7 @@ static long ucs_range[]={
0x1F0000, /* Reserved */
0x200000, /* Reserved */
0x210000, /* Reserved */
- 0x220000, /* Reserved 0xf8 */
+ 0x220000, /* Reserved 0x118 */
CJK_CHAR_LIMIT
};
@@ -352,10 +384,10 @@ integer kcatcodekey(integer c)
if ( (FULLWIDTH_DIGIT_0 <=c && c<=FULLWIDTH_DIGIT_9 )
|| (FULLWIDTH_CAPITAL_A<=c && c<=FULLWIDTH_CAPITAL_Z)
|| (FULLWIDTH_SMALL_A <=c && c<=FULLWIDTH_SMALL_Z ) )
- return 0xFE;
+ return 0x1FE;
if ( (HALFWIDTH_KATAKANA_WO <=c && c<=HALFWIDTH_KATAKANA_SMALL_TSU )
|| (HALFWIDTH_KATAKANA_A <=c && c<=HALFWIDTH_KATAKANA_N ) )
- return 0xFF;
+ return 0x1FF;
return binary_search((long)c, ucs_range, 0, NUCS_RANGE-1);
} else {
return Hi(toDVI(c));