diff options
-rw-r--r-- | ChangeLog | 8 | ||||
-rw-r--r-- | enc/unicode.c | 12 | ||||
-rwxr-xr-x | enc/unicode/case-folding.rb | 41 | ||||
-rw-r--r-- | enc/unicode/casefold.h | 445 |
4 files changed, 307 insertions, 199 deletions
@@ -1,3 +1,11 @@ +Fri Mar 11 16:11:27 2016 Martin Duerst <duerst@it.aoyama.ac.jp> + + * enc/unicode/case-folding.rb, casefold.h: Streamlining approach to + case mapping data not available from case folding by unifying all + three cases (special title, special upper, special lower). + * enc/unicode.c: Adjust macro names for above (macros are currently inactive). + (with Kimihito Matsui) + Thu Mar 10 17:34:16 2016 Nobuyoshi Nakada <nobu@ruby-lang.org> * iseq.c (prepare_iseq_build): enable coverage by coverage_enabled diff --git a/enc/unicode.c b/enc/unicode.c index e9c2803cab..f5a7d9b984 100644 --- a/enc/unicode.c +++ b/enc/unicode.c @@ -140,14 +140,22 @@ code3_equal(const OnigCodePoint *x, const OnigCodePoint *y) #define U ONIGENC_CASE_UPCASE #define D ONIGENC_CASE_DOWNCASE #define F ONIGENC_CASE_FOLD -#define T(n) (ONIGENC_CASE_TITLECASE|OnigTitlecaseEncode(n)) +#define ST 0 +#define SU 0 +#define SL 0 +#define I(n) 0 +#define L(n) 0 #include "enc/unicode/casefold.h" #undef U #undef D #undef F -#undef T +#undef ST +#undef SU +#undef SL +#undef I +#undef L #include "enc/unicode/name2ctype.h" diff --git a/enc/unicode/case-folding.rb b/enc/unicode/case-folding.rb index cd14b6e6f1..d309dd6d15 100755 --- a/enc/unicode/case-folding.rb +++ b/enc/unicode/case-folding.rb @@ -177,7 +177,7 @@ class CaseFolding dest.print lookup_hash(name, "CodePointList2", data) # TitleCase - dest.print mapping_data.titlecase_output + dest.print mapping_data.specials_output end def debug! @@ -203,7 +203,8 @@ end class CaseMapping def initialize (mapping_directory) @mappings = {} - @titlecase = [] + @specials = [] + @specials_length = 0 IO.readlines(File.expand_path('UnicodeData.txt', mapping_directory), encoding: Encoding::ASCII_8BIT).each do |line| next if line =~ /^</ code, _1,_2,_3,_4,_5,_6,_7,_8,_9,_10,_11, upper, lower, title = line.chomp.split ';' @@ -237,12 +238,24 @@ class CaseMapping if item flags += '|U' if to==item.upper flags += '|D' if to==item.lower + specials_index = nil + specials = [] unless item.upper == item.title - unless title_index = @titlecase.find_index { |i| i.title==item.title } - title_index = @titlecase.length - @titlecase << item - end - flags += "|T(#{title_index})" + specials << item.title + flags += "|ST" + end + unless item.lower.nil? or item.lower==from or item.lower==to + specials << item.lower + flags += "|SL" + end + unless item.upper.nil? or item.upper==from or item.upper==to + specials << item.upper + flags += "|SU" + end + if specials.first + flags += "|I(#{@specials_length})" + @specials_length += specials.map { |s| s.split(/ /).length }.reduce(:+) + @specials << specials end end flags @@ -252,12 +265,14 @@ class CaseMapping @debug = true end - def titlecase_output - "CodePointList3 TitleCase[] = {\n" + - @titlecase.map do |item| - chars = item.title.split(/ /) - ct = ' /* ' + Array(chars).map{|c|[c.to_i(16)].pack("U*")}.join(", ") + ' */' if @debug - " {#{chars.length}, {#{chars.map {|c| "0x"+c }.join(', ')}#{ct}}},\n" + def specials_output + "OnigCodePoint CaseMappingSpecials[] = {\n" + + @specials.map do |sps| + ' ' + sps.map do |sp| + chars = sp.split(/ /) + ct = ' /* ' + Array(chars).map{|c|[c.to_i(16)].pack("U*")}.join(", ") + ' */' if @debug + " L(#{chars.length})|#{chars.map {|c| "0x"+c }.join(', ')}#{ct}," + end.join + "\n" end.join + "};\n" end diff --git a/enc/unicode/casefold.h b/enc/unicode/casefold.h index 6e7140cf7f..fcc912035d 100644 --- a/enc/unicode/casefold.h +++ b/enc/unicode/casefold.h @@ -28,7 +28,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x0058, {1|F|D, {0x0078}}}, {0x0059, {1|F|D, {0x0079}}}, {0x005a, {1|F|D, {0x007a}}}, - {0x00b5, {1|F, {0x03bc}}}, + {0x00b5, {1|F|SU|I(0), {0x03bc}}}, {0x00c0, {1|F|D, {0x00e0}}}, {0x00c1, {1|F|D, {0x00e1}}}, {0x00c2, {1|F|D, {0x00e2}}}, @@ -59,7 +59,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x00dc, {1|F|D, {0x00fc}}}, {0x00dd, {1|F|D, {0x00fd}}}, {0x00de, {1|F|D, {0x00fe}}}, - {0x00df, {2|F|T(0), {0x0073, 0x0073}}}, + {0x00df, {2|F|ST|SU|I(1), {0x0073, 0x0073}}}, {0x0100, {1|F|D, {0x0101}}}, {0x0102, {1|F|D, {0x0103}}}, {0x0104, {1|F|D, {0x0105}}}, @@ -95,7 +95,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x0143, {1|F|D, {0x0144}}}, {0x0145, {1|F|D, {0x0146}}}, {0x0147, {1|F|D, {0x0148}}}, - {0x0149, {2|F, {0x02bc, 0x006e}}}, + {0x0149, {2|F|SU|I(5), {0x02bc, 0x006e}}}, {0x014a, {1|F|D, {0x014b}}}, {0x014c, {1|F|D, {0x014d}}}, {0x014e, {1|F|D, {0x014f}}}, @@ -123,7 +123,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x0179, {1|F|D, {0x017a}}}, {0x017b, {1|F|D, {0x017c}}}, {0x017d, {1|F|D, {0x017e}}}, - {0x017f, {1|F, {0x0073}}}, + {0x017f, {1|F|SU|I(7), {0x0073}}}, {0x0181, {1|F|D, {0x0253}}}, {0x0182, {1|F|D, {0x0183}}}, {0x0184, {1|F|D, {0x0185}}}, @@ -160,12 +160,12 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x01b7, {1|F|D, {0x0292}}}, {0x01b8, {1|F|D, {0x01b9}}}, {0x01bc, {1|F|D, {0x01bd}}}, - {0x01c4, {1|F|D|T(1), {0x01c6}}}, - {0x01c5, {1|F|D|T(1), {0x01c6}}}, - {0x01c7, {1|F|D|T(2), {0x01c9}}}, - {0x01c8, {1|F|D|T(2), {0x01c9}}}, - {0x01ca, {1|F|D|T(3), {0x01cc}}}, - {0x01cb, {1|F|D|T(3), {0x01cc}}}, + {0x01c4, {1|F|D|ST|I(8), {0x01c6}}}, + {0x01c5, {1|F|D|ST|SU|I(9), {0x01c6}}}, + {0x01c7, {1|F|D|ST|I(11), {0x01c9}}}, + {0x01c8, {1|F|D|ST|SU|I(12), {0x01c9}}}, + {0x01ca, {1|F|D|ST|I(14), {0x01cc}}}, + {0x01cb, {1|F|D|ST|SU|I(15), {0x01cc}}}, {0x01cd, {1|F|D, {0x01ce}}}, {0x01cf, {1|F|D, {0x01d0}}}, {0x01d1, {1|F|D, {0x01d2}}}, @@ -183,9 +183,9 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x01ea, {1|F|D, {0x01eb}}}, {0x01ec, {1|F|D, {0x01ed}}}, {0x01ee, {1|F|D, {0x01ef}}}, - {0x01f0, {2|F, {0x006a, 0x030c}}}, - {0x01f1, {1|F|D|T(4), {0x01f3}}}, - {0x01f2, {1|F|D|T(4), {0x01f3}}}, + {0x01f0, {2|F|SU|I(17), {0x006a, 0x030c}}}, + {0x01f1, {1|F|D|ST|I(19), {0x01f3}}}, + {0x01f2, {1|F|D|ST|SU|I(20), {0x01f3}}}, {0x01f4, {1|F|D, {0x01f5}}}, {0x01f6, {1|F|D, {0x0195}}}, {0x01f7, {1|F|D, {0x01bf}}}, @@ -232,7 +232,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x024a, {1|F|D, {0x024b}}}, {0x024c, {1|F|D, {0x024d}}}, {0x024e, {1|F|D, {0x024f}}}, - {0x0345, {1|F, {0x03b9}}}, + {0x0345, {1|F|SU|I(22), {0x03b9}}}, {0x0370, {1|F|D, {0x0371}}}, {0x0372, {1|F|D, {0x0373}}}, {0x0376, {1|F|D, {0x0377}}}, @@ -244,7 +244,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x038c, {1|F|D, {0x03cc}}}, {0x038e, {1|F|D, {0x03cd}}}, {0x038f, {1|F|D, {0x03ce}}}, - {0x0390, {3|F, {0x03b9, 0x0308, 0x0301}}}, + {0x0390, {3|F|SU|I(23), {0x03b9, 0x0308, 0x0301}}}, {0x0391, {1|F|D, {0x03b1}}}, {0x0392, {1|F|D, {0x03b2}}}, {0x0393, {1|F|D, {0x03b3}}}, @@ -271,13 +271,13 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x03a9, {1|F|D, {0x03c9}}}, {0x03aa, {1|F|D, {0x03ca}}}, {0x03ab, {1|F|D, {0x03cb}}}, - {0x03b0, {3|F, {0x03c5, 0x0308, 0x0301}}}, - {0x03c2, {1|F, {0x03c3}}}, + {0x03b0, {3|F|SU|I(26), {0x03c5, 0x0308, 0x0301}}}, + {0x03c2, {1|F|SU|I(29), {0x03c3}}}, {0x03cf, {1|F|D, {0x03d7}}}, - {0x03d0, {1|F, {0x03b2}}}, - {0x03d1, {1|F, {0x03b8}}}, - {0x03d5, {1|F, {0x03c6}}}, - {0x03d6, {1|F, {0x03c0}}}, + {0x03d0, {1|F|SU|I(30), {0x03b2}}}, + {0x03d1, {1|F|SU|I(31), {0x03b8}}}, + {0x03d5, {1|F|SU|I(32), {0x03c6}}}, + {0x03d6, {1|F|SU|I(33), {0x03c0}}}, {0x03d8, {1|F|D, {0x03d9}}}, {0x03da, {1|F|D, {0x03db}}}, {0x03dc, {1|F|D, {0x03dd}}}, @@ -290,10 +290,10 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x03ea, {1|F|D, {0x03eb}}}, {0x03ec, {1|F|D, {0x03ed}}}, {0x03ee, {1|F|D, {0x03ef}}}, - {0x03f0, {1|F, {0x03ba}}}, - {0x03f1, {1|F, {0x03c1}}}, + {0x03f0, {1|F|SU|I(34), {0x03ba}}}, + {0x03f1, {1|F|SU|I(35), {0x03c1}}}, {0x03f4, {1|F|D, {0x03b8}}}, - {0x03f5, {1|F, {0x03b5}}}, + {0x03f5, {1|F|SU|I(36), {0x03b5}}}, {0x03f7, {1|F|D, {0x03f8}}}, {0x03f9, {1|F|D, {0x03f2}}}, {0x03fa, {1|F|D, {0x03fb}}}, @@ -486,7 +486,7 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x0554, {1|F|D, {0x0584}}}, {0x0555, {1|F|D, {0x0585}}}, {0x0556, {1|F|D, {0x0586}}}, - {0x0587, {2|F|T(5), {0x0565, 0x0582}}}, + {0x0587, {2|F|ST|SU|I(37), {0x0565, 0x0582}}}, {0x10a0, {1|F|D, {0x2d00}}}, {0x10a1, {1|F|D, {0x2d01}}}, {0x10a2, {1|F|D, {0x2d02}}}, @@ -608,13 +608,13 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x1e90, {1|F|D, {0x1e91}}}, {0x1e92, {1|F|D, {0x1e93}}}, {0x1e94, {1|F|D, {0x1e95}}}, - {0x1e96, {2|F, {0x0068, 0x0331}}}, - {0x1e97, {2|F, {0x0074, 0x0308}}}, - {0x1e98, {2|F, {0x0077, 0x030a}}}, - {0x1e99, {2|F, {0x0079, 0x030a}}}, - {0x1e9a, {2|F, {0x0061, 0x02be}}}, - {0x1e9b, {1|F, {0x1e61}}}, - {0x1e9e, {2|F, {0x0073, 0x0073}}}, + {0x1e96, {2|F|SU|I(41), {0x0068, 0x0331}}}, + {0x1e97, {2|F|SU|I(43), {0x0074, 0x0308}}}, + {0x1e98, {2|F|SU|I(45), {0x0077, 0x030a}}}, + {0x1e99, {2|F|SU|I(47), {0x0079, 0x030a}}}, + {0x1e9a, {2|F|SU|I(49), {0x0061, 0x02be}}}, + {0x1e9b, {1|F|SU|I(51), {0x1e61}}}, + {0x1e9e, {2|F|SL|I(52), {0x0073, 0x0073}}}, {0x1ea0, {1|F|D, {0x1ea1}}}, {0x1ea2, {1|F|D, {0x1ea3}}}, {0x1ea4, {1|F|D, {0x1ea5}}}, @@ -699,10 +699,10 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x1f4b, {1|F|D, {0x1f43}}}, {0x1f4c, {1|F|D, {0x1f44}}}, {0x1f4d, {1|F|D, {0x1f45}}}, - {0x1f50, {2|F, {0x03c5, 0x0313}}}, - {0x1f52, {3|F, {0x03c5, 0x0313, 0x0300}}}, - {0x1f54, {3|F, {0x03c5, 0x0313, 0x0301}}}, - {0x1f56, {3|F, {0x03c5, 0x0313, 0x0342}}}, + {0x1f50, {2|F|SU|I(53), {0x03c5, 0x0313}}}, + {0x1f52, {3|F|SU|I(55), {0x03c5, 0x0313, 0x0300}}}, + {0x1f54, {3|F|SU|I(58), {0x03c5, 0x0313, 0x0301}}}, + {0x1f56, {3|F|SU|I(61), {0x03c5, 0x0313, 0x0342}}}, {0x1f59, {1|F|D, {0x1f51}}}, {0x1f5b, {1|F|D, {0x1f53}}}, {0x1f5d, {1|F|D, {0x1f55}}}, @@ -715,103 +715,103 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0x1f6d, {1|F|D, {0x1f65}}}, {0x1f6e, {1|F|D, {0x1f66}}}, {0x1f6f, {1|F|D, {0x1f67}}}, - {0x1f80, {2|F|T(6), {0x1f00, 0x03b9}}}, - {0x1f81, {2|F|T(7), {0x1f01, 0x03b9}}}, - {0x1f82, {2|F|T(8), {0x1f02, 0x03b9}}}, - {0x1f83, {2|F|T(9), {0x1f03, 0x03b9}}}, - {0x1f84, {2|F|T(10), {0x1f04, 0x03b9}}}, - {0x1f85, {2|F|T(11), {0x1f05, 0x03b9}}}, - {0x1f86, {2|F|T(12), {0x1f06, 0x03b9}}}, - {0x1f87, {2|F|T(13), {0x1f07, 0x03b9}}}, - {0x1f88, {2|F|T(6), {0x1f00, 0x03b9}}}, - {0x1f89, {2|F|T(7), {0x1f01, 0x03b9}}}, - {0x1f8a, {2|F|T(8), {0x1f02, 0x03b9}}}, - {0x1f8b, {2|F|T(9), {0x1f03, 0x03b9}}}, - {0x1f8c, {2|F|T(10), {0x1f04, 0x03b9}}}, - {0x1f8d, {2|F|T(11), {0x1f05, 0x03b9}}}, - {0x1f8e, {2|F|T(12), {0x1f06, 0x03b9}}}, - {0x1f8f, {2|F|T(13), {0x1f07, 0x03b9}}}, - {0x1f90, {2|F|T(14), {0x1f20, 0x03b9}}}, - {0x1f91, {2|F|T(15), {0x1f21, 0x03b9}}}, - {0x1f92, {2|F|T(16), {0x1f22, 0x03b9}}}, - {0x1f93, {2|F|T(17), {0x1f23, 0x03b9}}}, - {0x1f94, {2|F|T(18), {0x1f24, 0x03b9}}}, - {0x1f95, {2|F|T(19), {0x1f25, 0x03b9}}}, - {0x1f96, {2|F|T(20), {0x1f26, 0x03b9}}}, - {0x1f97, {2|F|T(21), {0x1f27, 0x03b9}}}, - {0x1f98, {2|F|T(14), {0x1f20, 0x03b9}}}, - {0x1f99, {2|F|T(15), {0x1f21, 0x03b9}}}, - {0x1f9a, {2|F|T(16), {0x1f22, 0x03b9}}}, - {0x1f9b, {2|F|T(17), {0x1f23, 0x03b9}}}, - {0x1f9c, {2|F|T(18), {0x1f24, 0x03b9}}}, - {0x1f9d, {2|F|T(19), {0x1f25, 0x03b9}}}, - {0x1f9e, {2|F|T(20), {0x1f26, 0x03b9}}}, - {0x1f9f, {2|F|T(21), {0x1f27, 0x03b9}}}, - {0x1fa0, {2|F|T(22), {0x1f60, 0x03b9}}}, - {0x1fa1, {2|F|T(23), {0x1f61, 0x03b9}}}, - {0x1fa2, {2|F|T(24), {0x1f62, 0x03b9}}}, - {0x1fa3, {2|F|T(25), {0x1f63, 0x03b9}}}, - {0x1fa4, {2|F|T(26), {0x1f64, 0x03b9}}}, - {0x1fa5, {2|F|T(27), {0x1f65, 0x03b9}}}, - {0x1fa6, {2|F|T(28), {0x1f66, 0x03b9}}}, - {0x1fa7, {2|F|T(29), {0x1f67, 0x03b9}}}, - {0x1fa8, {2|F|T(22), {0x1f60, 0x03b9}}}, - {0x1fa9, {2|F|T(23), {0x1f61, 0x03b9}}}, - {0x1faa, {2|F|T(24), {0x1f62, 0x03b9}}}, - {0x1fab, {2|F|T(25), {0x1f63, 0x03b9}}}, - {0x1fac, {2|F|T(26), {0x1f64, 0x03b9}}}, - {0x1fad, {2|F|T(27), {0x1f65, 0x03b9}}}, - {0x1fae, {2|F|T(28), {0x1f66, 0x03b9}}}, - {0x1faf, {2|F|T(29), {0x1f67, 0x03b9}}}, - {0x1fb2, {2|F|T(30), {0x1f70, 0x03b9}}}, - {0x1fb3, {2|F|T(31), {0x03b1, 0x03b9}}}, - {0x1fb4, {2|F|T(32), {0x03ac, 0x03b9}}}, - {0x1fb6, {2|F, {0x03b1, 0x0342}}}, - {0x1fb7, {3|F|T(33), {0x03b1, 0x0342, 0x03b9}}}, + {0x1f80, {2|F|ST|SU|I(64), {0x1f00, 0x03b9}}}, + {0x1f81, {2|F|ST|SU|I(67), {0x1f01, 0x03b9}}}, + {0x1f82, {2|F|ST|SU|I(70), {0x1f02, 0x03b9}}}, + {0x1f83, {2|F|ST|SU|I(73), {0x1f03, 0x03b9}}}, + {0x1f84, {2|F|ST|SU|I(76), {0x1f04, 0x03b9}}}, + {0x1f85, {2|F|ST|SU|I(79), {0x1f05, 0x03b9}}}, + {0x1f86, {2|F|ST|SU|I(82), {0x1f06, 0x03b9}}}, + {0x1f87, {2|F|ST|SU|I(85), {0x1f07, 0x03b9}}}, + {0x1f88, {2|F|ST|SL|SU|I(88), {0x1f00, 0x03b9}}}, + {0x1f89, {2|F|ST|SL|SU|I(92), {0x1f01, 0x03b9}}}, + {0x1f8a, {2|F|ST|SL|SU|I(96), {0x1f02, 0x03b9}}}, + {0x1f8b, {2|F|ST|SL|SU|I(100), {0x1f03, 0x03b9}}}, + {0x1f8c, {2|F|ST|SL|SU|I(104), {0x1f04, 0x03b9}}}, + {0x1f8d, {2|F|ST|SL|SU|I(108), {0x1f05, 0x03b9}}}, + {0x1f8e, {2|F|ST|SL|SU|I(112), {0x1f06, 0x03b9}}}, + {0x1f8f, {2|F|ST|SL|SU|I(116), {0x1f07, 0x03b9}}}, + {0x1f90, {2|F|ST|SU|I(120), {0x1f20, 0x03b9}}}, + {0x1f91, {2|F|ST|SU|I(123), {0x1f21, 0x03b9}}}, + {0x1f92, {2|F|ST|SU|I(126), {0x1f22, 0x03b9}}}, + {0x1f93, {2|F|ST|SU|I(129), {0x1f23, 0x03b9}}}, + {0x1f94, {2|F|ST|SU|I(132), {0x1f24, 0x03b9}}}, + {0x1f95, {2|F|ST|SU|I(135), {0x1f25, 0x03b9}}}, + {0x1f96, {2|F|ST|SU|I(138), {0x1f26, 0x03b9}}}, + {0x1f97, {2|F|ST|SU|I(141), {0x1f27, 0x03b9}}}, + {0x1f98, {2|F|ST|SL|SU|I(144), {0x1f20, 0x03b9}}}, + {0x1f99, {2|F|ST|SL|SU|I(148), {0x1f21, 0x03b9}}}, + {0x1f9a, {2|F|ST|SL|SU|I(152), {0x1f22, 0x03b9}}}, + {0x1f9b, {2|F|ST|SL|SU|I(156), {0x1f23, 0x03b9}}}, + {0x1f9c, {2|F|ST|SL|SU|I(160), {0x1f24, 0x03b9}}}, + {0x1f9d, {2|F|ST|SL|SU|I(164), {0x1f25, 0x03b9}}}, + {0x1f9e, {2|F|ST|SL|SU|I(168), {0x1f26, 0x03b9}}}, + {0x1f9f, {2|F|ST|SL|SU|I(172), {0x1f27, 0x03b9}}}, + {0x1fa0, {2|F|ST|SU|I(176), {0x1f60, 0x03b9}}}, + {0x1fa1, {2|F|ST|SU|I(179), {0x1f61, 0x03b9}}}, + {0x1fa2, {2|F|ST|SU|I(182), {0x1f62, 0x03b9}}}, + {0x1fa3, {2|F|ST|SU|I(185), {0x1f63, 0x03b9}}}, + {0x1fa4, {2|F|ST|SU|I(188), {0x1f64, 0x03b9}}}, + {0x1fa5, {2|F|ST|SU|I(191), {0x1f65, 0x03b9}}}, + {0x1fa6, {2|F|ST|SU|I(194), {0x1f66, 0x03b9}}}, + {0x1fa7, {2|F|ST|SU|I(197), {0x1f67, 0x03b9}}}, + {0x1fa8, {2|F|ST|SL|SU|I(200), {0x1f60, 0x03b9}}}, + {0x1fa9, {2|F|ST|SL|SU|I(204), {0x1f61, 0x03b9}}}, + {0x1faa, {2|F|ST|SL|SU|I(208), {0x1f62, 0x03b9}}}, + {0x1fab, {2|F|ST|SL|SU|I(212), {0x1f63, 0x03b9}}}, + {0x1fac, {2|F|ST|SL|SU|I(216), {0x1f64, 0x03b9}}}, + {0x1fad, {2|F|ST|SL|SU|I(220), {0x1f65, 0x03b9}}}, + {0x1fae, {2|F|ST|SL|SU|I(224), {0x1f66, 0x03b9}}}, + {0x1faf, {2|F|ST|SL|SU|I(228), {0x1f67, 0x03b9}}}, + {0x1fb2, {2|F|ST|SU|I(232), {0x1f70, 0x03b9}}}, + {0x1fb3, {2|F|ST|SU|I(236), {0x03b1, 0x03b9}}}, + {0x1fb4, {2|F|ST|SU|I(239), {0x03ac, 0x03b9}}}, + {0x1fb6, {2|F|SU|I(243), {0x03b1, 0x0342}}}, + {0x1fb7, {3|F|ST|SU|I(245), {0x03b1, 0x0342, 0x03b9}}}, {0x1fb8, {1|F|D, {0x1fb0}}}, {0x1fb9, {1|F|D, {0x1fb1}}}, {0x1fba, {1|F|D, {0x1f70}}}, {0x1fbb, {1|F|D, {0x1f71}}}, - {0x1fbc, {2|F|T(31), {0x03b1, 0x03b9}}}, - {0x1fbe, {1|F, {0x03b9}}}, - {0x1fc2, {2|F|T(34), {0x1f74, 0x03b9}}}, - {0x1fc3, {2|F|T(35), {0x03b7, 0x03b9}}}, - {0x1fc4, {2|F|T(36), {0x03ae, 0x03b9}}}, - {0x1fc6, {2|F, {0x03b7, 0x0342}}}, - {0x1fc7, {3|F|T(37), {0x03b7, 0x0342, 0x03b9}}}, + {0x1fbc, {2|F|ST|SL|SU|I(251), {0x03b1, 0x03b9}}}, + {0x1fbe, {1|F|SU|I(255), {0x03b9}}}, + {0x1fc2, {2|F|ST|SU|I(256), {0x1f74, 0x03b9}}}, + {0x1fc3, {2|F|ST|SU|I(260), {0x03b7, 0x03b9}}}, + {0x1fc4, {2|F|ST|SU|I(263), {0x03ae, 0x03b9}}}, + {0x1fc6, {2|F|SU|I(267), {0x03b7, 0x0342}}}, + {0x1fc7, {3|F|ST|SU|I(269), {0x03b7, 0x0342, 0x03b9}}}, {0x1fc8, {1|F|D, {0x1f72}}}, {0x1fc9, {1|F|D, {0x1f73}}}, {0x1fca, {1|F|D, {0x1f74}}}, {0x1fcb, {1|F|D, {0x1f75}}}, - {0x1fcc, {2|F|T(35), {0x03b7, 0x03b9}}}, - {0x1fd2, {3|F, {0x03b9, 0x0308, 0x0300}}}, - {0x1fd3, {3|F, {0x03b9, 0x0308, 0x0301}}}, - {0x1fd6, {2|F, {0x03b9, 0x0342}}}, - {0x1fd7, {3|F, {0x03b9, 0x0308, 0x0342}}}, + {0x1fcc, {2|F|ST|SL|SU|I(275), {0x03b7, 0x03b9}}}, + {0x1fd2, {3|F|SU|I(279), {0x03b9, 0x0308, 0x0300}}}, + {0x1fd3, {3|F|SU|I(282), {0x03b9, 0x0308, 0x0301}}}, + {0x1fd6, {2|F|SU|I(285), {0x03b9, 0x0342}}}, + {0x1fd7, {3|F|SU|I(287), {0x03b9, 0x0308, 0x0342}}}, {0x1fd8, {1|F|D, {0x1fd0}}}, {0x1fd9, {1|F|D, {0x1fd1}}}, {0x1fda, {1|F|D, {0x1f76}}}, {0x1fdb, {1|F|D, {0x1f77}}}, - {0x1fe2, {3|F, {0x03c5, 0x0308, 0x0300}}}, - {0x1fe3, {3|F, {0x03c5, 0x0308, 0x0301}}}, - {0x1fe4, {2|F, {0x03c1, 0x0313}}}, - {0x1fe6, {2|F, {0x03c5, 0x0342}}}, - {0x1fe7, {3|F, {0x03c5, 0x0308, 0x0342}}}, + {0x1fe2, {3|F|SU|I(290), {0x03c5, 0x0308, 0x0300}}}, + {0x1fe3, {3|F|SU|I(293), {0x03c5, 0x0308, 0x0301}}}, + {0x1fe4, {2|F|SU|I(296), {0x03c1, 0x0313}}}, + {0x1fe6, {2|F|SU|I(298), {0x03c5, 0x0342}}}, + {0x1fe7, {3|F|SU|I(300), {0x03c5, 0x0308, 0x0342}}}, {0x1fe8, {1|F|D, {0x1fe0}}}, {0x1fe9, {1|F|D, {0x1fe1}}}, {0x1fea, {1|F|D, {0x1f7a}}}, {0x1feb, {1|F|D, {0x1f7b}}}, {0x1fec, {1|F|D, {0x1fe5}}}, - {0x1ff2, {2|F|T(38), {0x1f7c, 0x03b9}}}, - {0x1ff3, {2|F|T(39), {0x03c9, 0x03b9}}}, - {0x1ff4, {2|F|T(40), {0x03ce, 0x03b9}}}, - {0x1ff6, {2|F, {0x03c9, 0x0342}}}, - {0x1ff7, {3|F|T(41), {0x03c9, 0x0342, 0x03b9}}}, + {0x1ff2, {2|F|ST|SU|I(303), {0x1f7c, 0x03b9}}}, + {0x1ff3, {2|F|ST|SU|I(307), {0x03c9, 0x03b9}}}, + {0x1ff4, {2|F|ST|SU|I(310), {0x03ce, 0x03b9}}}, + {0x1ff6, {2|F|SU|I(314), {0x03c9, 0x0342}}}, + {0x1ff7, {3|F|ST|SU|I(316), {0x03c9, 0x0342, 0x03b9}}}, {0x1ff8, {1|F|D, {0x1f78}}}, {0x1ff9, {1|F|D, {0x1f79}}}, {0x1ffa, {1|F|D, {0x1f7c}}}, {0x1ffb, {1|F|D, {0x1f7d}}}, - {0x1ffc, {2|F|T(39), {0x03c9, 0x03b9}}}, + {0x1ffc, {2|F|ST|SL|SU|I(322), {0x03c9, 0x03b9}}}, {0x2126, {1|F|D, {0x03c9}}}, {0x212a, {1|F|D, {0x006b}}}, {0x212b, {1|F|D, {0x00e5}}}, @@ -1161,18 +1161,18 @@ static const CaseFold_11_Type CaseFold_11_Table[] = { {0xabbd, {1|F|U, {0x13ed}}}, {0xabbe, {1|F|U, {0x13ee}}}, {0xabbf, {1|F|U, {0x13ef}}}, - {0xfb00, {2|F|T(42), {0x0066, 0x0066}}}, - {0xfb01, {2|F|T(43), {0x0066, 0x0069}}}, - {0xfb02, {2|F|T(44), {0x0066, 0x006c}}}, - {0xfb03, {3|F|T(45), {0x0066, 0x0066, 0x0069}}}, - {0xfb04, {3|F|T(46), {0x0066, 0x0066, 0x006c}}}, - {0xfb05, {2|F|T(47), {0x0073, 0x0074}}}, - {0xfb06, {2|F|T(47), {0x0073, 0x0074}}}, - {0xfb13, {2|F|T(48), {0x0574, 0x0576}}}, - {0xfb14, {2|F|T(49), {0x0574, 0x0565}}}, - {0xfb15, {2|F|T(50), {0x0574, 0x056b}}}, - {0xfb16, {2|F|T(51), {0x057e, 0x0576}}}, - {0xfb17, {2|F|T(52), {0x0574, 0x056d}}}, + {0xfb00, {2|F|ST|SU|I(326), {0x0066, 0x0066}}}, + {0xfb01, {2|F|ST|SU|I(330), {0x0066, 0x0069}}}, + {0xfb02, {2|F|ST|SU|I(334), {0x0066, 0x006c}}}, + {0xfb03, {3|F|ST|SU|I(338), {0x0066, 0x0066, 0x0069}}}, + {0xfb04, {3|F|ST|SU|I(344), {0x0066, 0x0066, 0x006c}}}, + {0xfb05, {2|F|ST|SU|I(350), {0x0073, 0x0074}}}, + {0xfb06, {2|F|ST|SU|I(354), {0x0073, 0x0074}}}, + {0xfb13, {2|F|ST|SU|I(358), {0x0574, 0x0576}}}, + {0xfb14, {2|F|ST|SU|I(362), {0x0574, 0x0565}}}, + {0xfb15, {2|F|ST|SU|I(366), {0x0574, 0x056b}}}, + {0xfb16, {2|F|ST|SU|I(370), {0x057e, 0x0576}}}, + {0xfb17, {2|F|ST|SU|I(374), {0x0574, 0x056d}}}, {0xff21, {1|F|D, {0xff41}}}, {0xff22, {1|F|D, {0xff42}}}, {0xff23, {1|F|D, {0xff43}}}, @@ -3298,9 +3298,9 @@ static const CaseUnfold_11_Type CaseUnfold_11_Table[] = { {0x01b9, {1|U, {0x01b8}}}, {0x01bd, {1|U, {0x01bc}}}, {0x01bf, {1|U, {0x01f7}}}, - {0x01c6, {2|U|T(1), {0x01c4, 0x01c5}}}, - {0x01c9, {2|U|T(2), {0x01c7, 0x01c8}}}, - {0x01cc, {2|U|T(3), {0x01ca, 0x01cb}}}, + {0x01c6, {2|U|ST|I(378), {0x01c4, 0x01c5}}}, + {0x01c9, {2|U|ST|I(379), {0x01c7, 0x01c8}}}, + {0x01cc, {2|U|ST|I(380), {0x01ca, 0x01cb}}}, {0x01ce, {1|U, {0x01cd}}}, {0x01d0, {1|U, {0x01cf}}}, {0x01d2, {1|U, {0x01d1}}}, @@ -3319,7 +3319,7 @@ static const CaseUnfold_11_Type CaseUnfold_11_Table[] = { {0x01eb, {1|U, {0x01ea}}}, {0x01ed, {1|U, {0x01ec}}}, {0x01ef, {1|U, {0x01ee}}}, - {0x01f3, {2|U|T(4), {0x01f1, 0x01f2}}}, + {0x01f3, {2|U|ST|I(381), {0x01f1, 0x01f2}}}, {0x01f5, {1|U, {0x01f4}}}, {0x01f9, {1|U, {0x01f8}}}, {0x01fb, {1|U, {0x01fa}}}, @@ -3412,10 +3412,10 @@ static const CaseUnfold_11_Type CaseUnfold_11_Table[] = { {0x03b6, {1|U, {0x0396}}}, {0x03b7, {1|U, {0x0397}}}, {0x03b8, {3|U, {0x0398, 0x03d1, 0x03f4}}}, - {0x03b9, {3, {0x0345, 0x0399, 0x1fbe}}}, + {0x03b9, {3|SU|I(382), {0x0345, 0x0399, 0x1fbe}}}, {0x03ba, {2|U, {0x039a, 0x03f0}}}, {0x03bb, {1|U, {0x039b}}}, - {0x03bc, {2, {0x00b5, 0x039c}}}, + {0x03bc, {2|SU|I(383), {0x00b5, 0x039c}}}, {0x03bd, {1|U, {0x039d}}}, {0x03be, {1|U, {0x039e}}}, {0x03bf, {1|U, {0x039f}}}, @@ -6246,58 +6246,135 @@ onigenc_unicode_CaseUnfold_13_lookup(const OnigCodePoint *codes) return 0; } -CodePointList3 TitleCase[] = { - {2, {0x0053, 0x0073}}, - {1, {0x01C5}}, - {1, {0x01C8}}, - {1, {0x01CB}}, - {1, {0x01F2}}, - {2, {0x0535, 0x0582}}, - {1, {0x1F88}}, - {1, {0x1F89}}, - {1, {0x1F8A}}, - {1, {0x1F8B}}, - {1, {0x1F8C}}, - {1, {0x1F8D}}, - {1, {0x1F8E}}, - {1, {0x1F8F}}, - {1, {0x1F98}}, - {1, {0x1F99}}, - {1, {0x1F9A}}, - {1, {0x1F9B}}, - {1, {0x1F9C}}, - {1, {0x1F9D}}, - {1, {0x1F9E}}, - {1, {0x1F9F}}, - {1, {0x1FA8}}, - {1, {0x1FA9}}, - {1, {0x1FAA}}, - {1, {0x1FAB}}, - {1, {0x1FAC}}, - {1, {0x1FAD}}, - {1, {0x1FAE}}, - {1, {0x1FAF}}, - {2, {0x1FBA, 0x0345}}, - {1, {0x1FBC}}, - {2, {0x0386, 0x0345}}, - {3, {0x0391, 0x0342, 0x0345}}, - {2, {0x1FCA, 0x0345}}, - {1, {0x1FCC}}, - {2, {0x0389, 0x0345}}, - {3, {0x0397, 0x0342, 0x0345}}, - {2, {0x1FFA, 0x0345}}, - {1, {0x1FFC}}, - {2, {0x038F, 0x0345}}, - {3, {0x03A9, 0x0342, 0x0345}}, - {2, {0x0046, 0x0066}}, - {2, {0x0046, 0x0069}}, - {2, {0x0046, 0x006C}}, - {3, {0x0046, 0x0066, 0x0069}}, - {3, {0x0046, 0x0066, 0x006C}}, - {2, {0x0053, 0x0074}}, - {2, {0x0544, 0x0576}}, - {2, {0x0544, 0x0565}}, - {2, {0x0544, 0x056B}}, - {2, {0x054E, 0x0576}}, - {2, {0x0544, 0x056D}}, +OnigCodePoint CaseMappingSpecials[] = { + L(1)|0x039C, + L(2)|0x0053, 0x0073, L(2)|0x0053, 0x0053, + L(2)|0x02BC, 0x004E, + L(1)|0x0053, + L(1)|0x01C5, + L(1)|0x01C5, L(1)|0x01C4, + L(1)|0x01C8, + L(1)|0x01C8, L(1)|0x01C7, + L(1)|0x01CB, + L(1)|0x01CB, L(1)|0x01CA, + L(2)|0x004A, 0x030C, + L(1)|0x01F2, + L(1)|0x01F2, L(1)|0x01F1, + L(1)|0x0399, + L(3)|0x0399, 0x0308, 0x0301, + L(3)|0x03A5, 0x0308, 0x0301, + L(1)|0x03A3, + L(1)|0x0392, + L(1)|0x0398, + L(1)|0x03A6, + L(1)|0x03A0, + L(1)|0x039A, + L(1)|0x03A1, + L(1)|0x0395, + L(2)|0x0535, 0x0582, L(2)|0x0535, 0x0552, + L(2)|0x0048, 0x0331, + L(2)|0x0054, 0x0308, + L(2)|0x0057, 0x030A, + L(2)|0x0059, 0x030A, + L(2)|0x0041, 0x02BE, + L(1)|0x1E60, + L(1)|0x00DF, + L(2)|0x03A5, 0x0313, + L(3)|0x03A5, 0x0313, 0x0300, + L(3)|0x03A5, 0x0313, 0x0301, + L(3)|0x03A5, 0x0313, 0x0342, + L(1)|0x1F88, L(2)|0x1F08, 0x0399, + L(1)|0x1F89, L(2)|0x1F09, 0x0399, + L(1)|0x1F8A, L(2)|0x1F0A, 0x0399, + L(1)|0x1F8B, L(2)|0x1F0B, 0x0399, + L(1)|0x1F8C, L(2)|0x1F0C, 0x0399, + L(1)|0x1F8D, L(2)|0x1F0D, 0x0399, + L(1)|0x1F8E, L(2)|0x1F0E, 0x0399, + L(1)|0x1F8F, L(2)|0x1F0F, 0x0399, + L(1)|0x1F88, L(1)|0x1F80, L(2)|0x1F08, 0x0399, + L(1)|0x1F89, L(1)|0x1F81, L(2)|0x1F09, 0x0399, + L(1)|0x1F8A, L(1)|0x1F82, L(2)|0x1F0A, 0x0399, + L(1)|0x1F8B, L(1)|0x1F83, L(2)|0x1F0B, 0x0399, + L(1)|0x1F8C, L(1)|0x1F84, L(2)|0x1F0C, 0x0399, + L(1)|0x1F8D, L(1)|0x1F85, L(2)|0x1F0D, 0x0399, + L(1)|0x1F8E, L(1)|0x1F86, L(2)|0x1F0E, 0x0399, + L(1)|0x1F8F, L(1)|0x1F87, L(2)|0x1F0F, 0x0399, + L(1)|0x1F98, L(2)|0x1F28, 0x0399, + L(1)|0x1F99, L(2)|0x1F29, 0x0399, + L(1)|0x1F9A, L(2)|0x1F2A, 0x0399, + L(1)|0x1F9B, L(2)|0x1F2B, 0x0399, + L(1)|0x1F9C, L(2)|0x1F2C, 0x0399, + L(1)|0x1F9D, L(2)|0x1F2D, 0x0399, + L(1)|0x1F9E, L(2)|0x1F2E, 0x0399, + L(1)|0x1F9F, L(2)|0x1F2F, 0x0399, + L(1)|0x1F98, L(1)|0x1F90, L(2)|0x1F28, 0x0399, + L(1)|0x1F99, L(1)|0x1F91, L(2)|0x1F29, 0x0399, + L(1)|0x1F9A, L(1)|0x1F92, L(2)|0x1F2A, 0x0399, + L(1)|0x1F9B, L(1)|0x1F93, L(2)|0x1F2B, 0x0399, + L(1)|0x1F9C, L(1)|0x1F94, L(2)|0x1F2C, 0x0399, + L(1)|0x1F9D, L(1)|0x1F95, L(2)|0x1F2D, 0x0399, + L(1)|0x1F9E, L(1)|0x1F96, L(2)|0x1F2E, 0x0399, + L(1)|0x1F9F, L(1)|0x1F97, L(2)|0x1F2F, 0x0399, + L(1)|0x1FA8, L(2)|0x1F68, 0x0399, + L(1)|0x1FA9, L(2)|0x1F69, 0x0399, + L(1)|0x1FAA, L(2)|0x1F6A, 0x0399, + L(1)|0x1FAB, L(2)|0x1F6B, 0x0399, + L(1)|0x1FAC, L(2)|0x1F6C, 0x0399, + L(1)|0x1FAD, L(2)|0x1F6D, 0x0399, + L(1)|0x1FAE, L(2)|0x1F6E, 0x0399, + L(1)|0x1FAF, L(2)|0x1F6F, 0x0399, + L(1)|0x1FA8, L(1)|0x1FA0, L(2)|0x1F68, 0x0399, + L(1)|0x1FA9, L(1)|0x1FA1, L(2)|0x1F69, 0x0399, + L(1)|0x1FAA, L(1)|0x1FA2, L(2)|0x1F6A, 0x0399, + L(1)|0x1FAB, L(1)|0x1FA3, L(2)|0x1F6B, 0x0399, + L(1)|0x1FAC, L(1)|0x1FA4, L(2)|0x1F6C, 0x0399, + L(1)|0x1FAD, L(1)|0x1FA5, L(2)|0x1F6D, 0x0399, + L(1)|0x1FAE, L(1)|0x1FA6, L(2)|0x1F6E, 0x0399, + L(1)|0x1FAF, L(1)|0x1FA7, L(2)|0x1F6F, 0x0399, + L(2)|0x1FBA, 0x0345, L(2)|0x1FBA, 0x0399, + L(1)|0x1FBC, L(2)|0x0391, 0x0399, + L(2)|0x0386, 0x0345, L(2)|0x0386, 0x0399, + L(2)|0x0391, 0x0342, + L(3)|0x0391, 0x0342, 0x0345, L(3)|0x0391, 0x0342, 0x0399, + L(1)|0x1FBC, L(1)|0x1FB3, L(2)|0x0391, 0x0399, + L(1)|0x0399, + L(2)|0x1FCA, 0x0345, L(2)|0x1FCA, 0x0399, + L(1)|0x1FCC, L(2)|0x0397, 0x0399, + L(2)|0x0389, 0x0345, L(2)|0x0389, 0x0399, + L(2)|0x0397, 0x0342, + L(3)|0x0397, 0x0342, 0x0345, L(3)|0x0397, 0x0342, 0x0399, + L(1)|0x1FCC, L(1)|0x1FC3, L(2)|0x0397, 0x0399, + L(3)|0x0399, 0x0308, 0x0300, + L(3)|0x0399, 0x0308, 0x0301, + L(2)|0x0399, 0x0342, + L(3)|0x0399, 0x0308, 0x0342, + L(3)|0x03A5, 0x0308, 0x0300, + L(3)|0x03A5, 0x0308, 0x0301, + L(2)|0x03A1, 0x0313, + L(2)|0x03A5, 0x0342, + L(3)|0x03A5, 0x0308, 0x0342, + L(2)|0x1FFA, 0x0345, L(2)|0x1FFA, 0x0399, + L(1)|0x1FFC, L(2)|0x03A9, 0x0399, + L(2)|0x038F, 0x0345, L(2)|0x038F, 0x0399, + L(2)|0x03A9, 0x0342, + L(3)|0x03A9, 0x0342, 0x0345, L(3)|0x03A9, 0x0342, 0x0399, + L(1)|0x1FFC, L(1)|0x1FF3, L(2)|0x03A9, 0x0399, + L(2)|0x0046, 0x0066, L(2)|0x0046, 0x0046, + L(2)|0x0046, 0x0069, L(2)|0x0046, 0x0049, + L(2)|0x0046, 0x006C, L(2)|0x0046, 0x004C, + L(3)|0x0046, 0x0066, 0x0069, L(3)|0x0046, 0x0046, 0x0049, + L(3)|0x0046, 0x0066, 0x006C, L(3)|0x0046, 0x0046, 0x004C, + L(2)|0x0053, 0x0074, L(2)|0x0053, 0x0054, + L(2)|0x0053, 0x0074, L(2)|0x0053, 0x0054, + L(2)|0x0544, 0x0576, L(2)|0x0544, 0x0546, + L(2)|0x0544, 0x0565, L(2)|0x0544, 0x0535, + L(2)|0x0544, 0x056B, L(2)|0x0544, 0x053B, + L(2)|0x054E, 0x0576, L(2)|0x054E, 0x0546, + L(2)|0x0544, 0x056D, L(2)|0x0544, 0x053D, + L(1)|0x01C5, + L(1)|0x01C8, + L(1)|0x01CB, + L(1)|0x01F2, + L(1)|0x0399, + L(1)|0x039C, }; |