summaryrefslogtreecommitdiff
path: root/src/unicode
diff options
context:
space:
mode:
authorRuss Cox <rsc@golang.org>2014-09-08 00:08:51 -0400
committerRuss Cox <rsc@golang.org>2014-09-08 00:08:51 -0400
commit8528da672cc093d4dd06732819abc1f7b6b5a46e (patch)
tree334be80d4a4c85b77db6f6fdb67cbf0528cba5f5 /src/unicode
parent73bcb69f272cbf34ddcc9daa56427a8683b5a95d (diff)
downloadgo-8528da672cc093d4dd06732819abc1f7b6b5a46e.tar.gz
build: move package sources from src/pkg to src
Preparation was in CL 134570043. This CL contains only the effect of 'hg mv src/pkg/* src'. For more about the move, see golang.org/s/go14nopkg.
Diffstat (limited to 'src/unicode')
-rw-r--r--src/unicode/casetables.go20
-rw-r--r--src/unicode/digit.go13
-rw-r--r--src/unicode/digit_test.go126
-rw-r--r--src/unicode/graphic.go144
-rw-r--r--src/unicode/graphic_test.go122
-rw-r--r--src/unicode/letter.go357
-rw-r--r--src/unicode/letter_test.go535
-rw-r--r--src/unicode/maketables.go1376
-rw-r--r--src/unicode/script_test.go287
-rw-r--r--src/unicode/tables.go7029
-rw-r--r--src/unicode/utf16/export_test.go11
-rw-r--r--src/unicode/utf16/utf16.go108
-rw-r--r--src/unicode/utf16/utf16_test.go149
-rw-r--r--src/unicode/utf8/example_test.go196
-rw-r--r--src/unicode/utf8/utf8.go435
-rw-r--r--src/unicode/utf8/utf8_test.go444
16 files changed, 11352 insertions, 0 deletions
diff --git a/src/unicode/casetables.go b/src/unicode/casetables.go
new file mode 100644
index 000000000..29bf167e5
--- /dev/null
+++ b/src/unicode/casetables.go
@@ -0,0 +1,20 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// TODO: This file contains the special casing rules for Turkish and Azeri only.
+// It should encompass all the languages with special casing rules
+// and be generated automatically, but that requires some API
+// development first.
+
+package unicode
+
+var TurkishCase SpecialCase = _TurkishCase
+var _TurkishCase = SpecialCase{
+ CaseRange{0x0049, 0x0049, d{0, 0x131 - 0x49, 0}},
+ CaseRange{0x0069, 0x0069, d{0x130 - 0x69, 0, 0x130 - 0x69}},
+ CaseRange{0x0130, 0x0130, d{0, 0x69 - 0x130, 0}},
+ CaseRange{0x0131, 0x0131, d{0x49 - 0x131, 0, 0x49 - 0x131}},
+}
+
+var AzeriCase SpecialCase = _TurkishCase
diff --git a/src/unicode/digit.go b/src/unicode/digit.go
new file mode 100644
index 000000000..53171b396
--- /dev/null
+++ b/src/unicode/digit.go
@@ -0,0 +1,13 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode
+
+// IsDigit reports whether the rune is a decimal digit.
+func IsDigit(r rune) bool {
+ if r <= MaxLatin1 {
+ return '0' <= r && r <= '9'
+ }
+ return isExcludingLatin(Digit, r)
+}
diff --git a/src/unicode/digit_test.go b/src/unicode/digit_test.go
new file mode 100644
index 000000000..551c42a24
--- /dev/null
+++ b/src/unicode/digit_test.go
@@ -0,0 +1,126 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode_test
+
+import (
+ "testing"
+ . "unicode"
+)
+
+var testDigit = []rune{
+ 0x0030,
+ 0x0039,
+ 0x0661,
+ 0x06F1,
+ 0x07C9,
+ 0x0966,
+ 0x09EF,
+ 0x0A66,
+ 0x0AEF,
+ 0x0B66,
+ 0x0B6F,
+ 0x0BE6,
+ 0x0BEF,
+ 0x0C66,
+ 0x0CEF,
+ 0x0D66,
+ 0x0D6F,
+ 0x0E50,
+ 0x0E59,
+ 0x0ED0,
+ 0x0ED9,
+ 0x0F20,
+ 0x0F29,
+ 0x1040,
+ 0x1049,
+ 0x1090,
+ 0x1091,
+ 0x1099,
+ 0x17E0,
+ 0x17E9,
+ 0x1810,
+ 0x1819,
+ 0x1946,
+ 0x194F,
+ 0x19D0,
+ 0x19D9,
+ 0x1B50,
+ 0x1B59,
+ 0x1BB0,
+ 0x1BB9,
+ 0x1C40,
+ 0x1C49,
+ 0x1C50,
+ 0x1C59,
+ 0xA620,
+ 0xA629,
+ 0xA8D0,
+ 0xA8D9,
+ 0xA900,
+ 0xA909,
+ 0xAA50,
+ 0xAA59,
+ 0xFF10,
+ 0xFF19,
+ 0x104A1,
+ 0x1D7CE,
+}
+
+var testLetter = []rune{
+ 0x0041,
+ 0x0061,
+ 0x00AA,
+ 0x00BA,
+ 0x00C8,
+ 0x00DB,
+ 0x00F9,
+ 0x02EC,
+ 0x0535,
+ 0x06E6,
+ 0x093D,
+ 0x0A15,
+ 0x0B99,
+ 0x0DC0,
+ 0x0EDD,
+ 0x1000,
+ 0x1200,
+ 0x1312,
+ 0x1401,
+ 0x1885,
+ 0x2C00,
+ 0xA800,
+ 0xF900,
+ 0xFA30,
+ 0xFFDA,
+ 0xFFDC,
+ 0x10000,
+ 0x10300,
+ 0x10400,
+ 0x20000,
+ 0x2F800,
+ 0x2FA1D,
+}
+
+func TestDigit(t *testing.T) {
+ for _, r := range testDigit {
+ if !IsDigit(r) {
+ t.Errorf("IsDigit(U+%04X) = false, want true", r)
+ }
+ }
+ for _, r := range testLetter {
+ if IsDigit(r) {
+ t.Errorf("IsDigit(U+%04X) = true, want false", r)
+ }
+ }
+}
+
+// Test that the special case in IsDigit agrees with the table
+func TestDigitOptimization(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ if Is(Digit, i) != IsDigit(i) {
+ t.Errorf("IsDigit(U+%04X) disagrees with Is(Digit)", i)
+ }
+ }
+}
diff --git a/src/unicode/graphic.go b/src/unicode/graphic.go
new file mode 100644
index 000000000..ba90b4e51
--- /dev/null
+++ b/src/unicode/graphic.go
@@ -0,0 +1,144 @@
+// Copyright 2011 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode
+
+// Bit masks for each code point under U+0100, for fast lookup.
+const (
+ pC = 1 << iota // a control character.
+ pP // a punctuation character.
+ pN // a numeral.
+ pS // a symbolic character.
+ pZ // a spacing character.
+ pLu // an upper-case letter.
+ pLl // a lower-case letter.
+ pp // a printable character according to Go's definition.
+ pg = pp | pZ // a graphical character according to the Unicode definition.
+ pLo = pLl | pLu // a letter that is neither upper nor lower case.
+ pLmask = pLo
+)
+
+// GraphicRanges defines the set of graphic characters according to Unicode.
+var GraphicRanges = []*RangeTable{
+ L, M, N, P, S, Zs,
+}
+
+// PrintRanges defines the set of printable characters according to Go.
+// ASCII space, U+0020, is handled separately.
+var PrintRanges = []*RangeTable{
+ L, M, N, P, S,
+}
+
+// IsGraphic reports whether the rune is defined as a Graphic by Unicode.
+// Such characters include letters, marks, numbers, punctuation, symbols, and
+// spaces, from categories L, M, N, P, S, Zs.
+func IsGraphic(r rune) bool {
+ // We convert to uint32 to avoid the extra test for negative,
+ // and in the index we convert to uint8 to avoid the range check.
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pg != 0
+ }
+ return In(r, GraphicRanges...)
+}
+
+// IsPrint reports whether the rune is defined as printable by Go. Such
+// characters include letters, marks, numbers, punctuation, symbols, and the
+// ASCII space character, from categories L, M, N, P, S and the ASCII space
+// character. This categorization is the same as IsGraphic except that the
+// only spacing character is ASCII space, U+0020.
+func IsPrint(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pp != 0
+ }
+ return In(r, PrintRanges...)
+}
+
+// IsOneOf reports whether the rune is a member of one of the ranges.
+// The function "In" provides a nicer signature and should be used in preference to IsOneOf.
+func IsOneOf(ranges []*RangeTable, r rune) bool {
+ for _, inside := range ranges {
+ if Is(inside, r) {
+ return true
+ }
+ }
+ return false
+}
+
+// In reports whether the rune is a member of one of the ranges.
+func In(r rune, ranges ...*RangeTable) bool {
+ for _, inside := range ranges {
+ if Is(inside, r) {
+ return true
+ }
+ }
+ return false
+}
+
+// IsControl reports whether the rune is a control character.
+// The C (Other) Unicode category includes more code points
+// such as surrogates; use Is(C, r) to test for them.
+func IsControl(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pC != 0
+ }
+ // All control characters are < Latin1Max.
+ return false
+}
+
+// IsLetter reports whether the rune is a letter (category L).
+func IsLetter(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&(pLmask) != 0
+ }
+ return isExcludingLatin(Letter, r)
+}
+
+// IsMark reports whether the rune is a mark character (category M).
+func IsMark(r rune) bool {
+ // There are no mark characters in Latin-1.
+ return isExcludingLatin(Mark, r)
+}
+
+// IsNumber reports whether the rune is a number (category N).
+func IsNumber(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pN != 0
+ }
+ return isExcludingLatin(Number, r)
+}
+
+// IsPunct reports whether the rune is a Unicode punctuation character
+// (category P).
+func IsPunct(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pP != 0
+ }
+ return Is(Punct, r)
+}
+
+// IsSpace reports whether the rune is a space character as defined
+// by Unicode's White Space property; in the Latin-1 space
+// this is
+// '\t', '\n', '\v', '\f', '\r', ' ', U+0085 (NEL), U+00A0 (NBSP).
+// Other definitions of spacing characters are set by category
+// Z and property Pattern_White_Space.
+func IsSpace(r rune) bool {
+ // This property isn't the same as Z; special-case it.
+ if uint32(r) <= MaxLatin1 {
+ switch r {
+ case '\t', '\n', '\v', '\f', '\r', ' ', 0x85, 0xA0:
+ return true
+ }
+ return false
+ }
+ return isExcludingLatin(White_Space, r)
+}
+
+// IsSymbol reports whether the rune is a symbolic character.
+func IsSymbol(r rune) bool {
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pS != 0
+ }
+ return isExcludingLatin(Symbol, r)
+}
diff --git a/src/unicode/graphic_test.go b/src/unicode/graphic_test.go
new file mode 100644
index 000000000..c9f289c7f
--- /dev/null
+++ b/src/unicode/graphic_test.go
@@ -0,0 +1,122 @@
+// Copyright 2011 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode_test
+
+import (
+ "testing"
+ . "unicode"
+)
+
+// Independently check that the special "Is" functions work
+// in the Latin-1 range through the property table.
+
+func TestIsControlLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsControl(i)
+ want := false
+ switch {
+ case 0x00 <= i && i <= 0x1F:
+ want = true
+ case 0x7F <= i && i <= 0x9F:
+ want = true
+ }
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsLetterLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsLetter(i)
+ want := Is(Letter, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsUpperLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsUpper(i)
+ want := Is(Upper, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsLowerLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsLower(i)
+ want := Is(Lower, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestNumberLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsNumber(i)
+ want := Is(Number, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsPrintLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsPrint(i)
+ want := In(i, PrintRanges...)
+ if i == ' ' {
+ want = true
+ }
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsGraphicLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsGraphic(i)
+ want := In(i, GraphicRanges...)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsPunctLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsPunct(i)
+ want := Is(Punct, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsSpaceLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsSpace(i)
+ want := Is(White_Space, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
+
+func TestIsSymbolLatin1(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ got := IsSymbol(i)
+ want := Is(Symbol, i)
+ if got != want {
+ t.Errorf("%U incorrect: got %t; want %t", i, got, want)
+ }
+ }
+}
diff --git a/src/unicode/letter.go b/src/unicode/letter.go
new file mode 100644
index 000000000..7fe4241ed
--- /dev/null
+++ b/src/unicode/letter.go
@@ -0,0 +1,357 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package unicode provides data and functions to test some properties of
+// Unicode code points.
+package unicode
+
+// Tables are regenerated each time we update the Unicode version.
+//go:generate go run maketables.go -tables=all -output tables.go
+
+const (
+ MaxRune = '\U0010FFFF' // Maximum valid Unicode code point.
+ ReplacementChar = '\uFFFD' // Represents invalid code points.
+ MaxASCII = '\u007F' // maximum ASCII value.
+ MaxLatin1 = '\u00FF' // maximum Latin-1 value.
+)
+
+// RangeTable defines a set of Unicode code points by listing the ranges of
+// code points within the set. The ranges are listed in two slices
+// to save space: a slice of 16-bit ranges and a slice of 32-bit ranges.
+// The two slices must be in sorted order and non-overlapping.
+// Also, R32 should contain only values >= 0x10000 (1<<16).
+type RangeTable struct {
+ R16 []Range16
+ R32 []Range32
+ LatinOffset int // number of entries in R16 with Hi <= MaxLatin1
+}
+
+// Range16 represents of a range of 16-bit Unicode code points. The range runs from Lo to Hi
+// inclusive and has the specified stride.
+type Range16 struct {
+ Lo uint16
+ Hi uint16
+ Stride uint16
+}
+
+// Range32 represents of a range of Unicode code points and is used when one or
+// more of the values will not fit in 16 bits. The range runs from Lo to Hi
+// inclusive and has the specified stride. Lo and Hi must always be >= 1<<16.
+type Range32 struct {
+ Lo uint32
+ Hi uint32
+ Stride uint32
+}
+
+// CaseRange represents a range of Unicode code points for simple (one
+// code point to one code point) case conversion.
+// The range runs from Lo to Hi inclusive, with a fixed stride of 1. Deltas
+// are the number to add to the code point to reach the code point for a
+// different case for that character. They may be negative. If zero, it
+// means the character is in the corresponding case. There is a special
+// case representing sequences of alternating corresponding Upper and Lower
+// pairs. It appears with a fixed Delta of
+// {UpperLower, UpperLower, UpperLower}
+// The constant UpperLower has an otherwise impossible delta value.
+type CaseRange struct {
+ Lo uint32
+ Hi uint32
+ Delta d
+}
+
+// SpecialCase represents language-specific case mappings such as Turkish.
+// Methods of SpecialCase customize (by overriding) the standard mappings.
+type SpecialCase []CaseRange
+
+// BUG(r): There is no mechanism for full case folding, that is, for
+// characters that involve multiple runes in the input or output.
+
+// Indices into the Delta arrays inside CaseRanges for case mapping.
+const (
+ UpperCase = iota
+ LowerCase
+ TitleCase
+ MaxCase
+)
+
+type d [MaxCase]rune // to make the CaseRanges text shorter
+
+// If the Delta field of a CaseRange is UpperLower, it means
+// this CaseRange represents a sequence of the form (say)
+// Upper Lower Upper Lower.
+const (
+ UpperLower = MaxRune + 1 // (Cannot be a valid delta.)
+)
+
+// linearMax is the maximum size table for linear search for non-Latin1 rune.
+// Derived by running 'go test -calibrate'.
+const linearMax = 18
+
+// is16 reports whether r is in the sorted slice of 16-bit ranges.
+func is16(ranges []Range16, r uint16) bool {
+ if len(ranges) <= linearMax || r <= MaxLatin1 {
+ for i := range ranges {
+ range_ := &ranges[i]
+ if r < range_.Lo {
+ return false
+ }
+ if r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ }
+ return false
+ }
+
+ // binary search over ranges
+ lo := 0
+ hi := len(ranges)
+ for lo < hi {
+ m := lo + (hi-lo)/2
+ range_ := &ranges[m]
+ if range_.Lo <= r && r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ if r < range_.Lo {
+ hi = m
+ } else {
+ lo = m + 1
+ }
+ }
+ return false
+}
+
+// is32 reports whether r is in the sorted slice of 32-bit ranges.
+func is32(ranges []Range32, r uint32) bool {
+ if len(ranges) <= linearMax {
+ for i := range ranges {
+ range_ := &ranges[i]
+ if r < range_.Lo {
+ return false
+ }
+ if r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ }
+ return false
+ }
+
+ // binary search over ranges
+ lo := 0
+ hi := len(ranges)
+ for lo < hi {
+ m := lo + (hi-lo)/2
+ range_ := ranges[m]
+ if range_.Lo <= r && r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ if r < range_.Lo {
+ hi = m
+ } else {
+ lo = m + 1
+ }
+ }
+ return false
+}
+
+// Is reports whether the rune is in the specified table of ranges.
+func Is(rangeTab *RangeTable, r rune) bool {
+ r16 := rangeTab.R16
+ if len(r16) > 0 && r <= rune(r16[len(r16)-1].Hi) {
+ return is16(r16, uint16(r))
+ }
+ r32 := rangeTab.R32
+ if len(r32) > 0 && r >= rune(r32[0].Lo) {
+ return is32(r32, uint32(r))
+ }
+ return false
+}
+
+func isExcludingLatin(rangeTab *RangeTable, r rune) bool {
+ r16 := rangeTab.R16
+ if off := rangeTab.LatinOffset; len(r16) > off && r <= rune(r16[len(r16)-1].Hi) {
+ return is16(r16[off:], uint16(r))
+ }
+ r32 := rangeTab.R32
+ if len(r32) > 0 && r >= rune(r32[0].Lo) {
+ return is32(r32, uint32(r))
+ }
+ return false
+}
+
+// IsUpper reports whether the rune is an upper case letter.
+func IsUpper(r rune) bool {
+ // See comment in IsGraphic.
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pLmask == pLu
+ }
+ return isExcludingLatin(Upper, r)
+}
+
+// IsLower reports whether the rune is a lower case letter.
+func IsLower(r rune) bool {
+ // See comment in IsGraphic.
+ if uint32(r) <= MaxLatin1 {
+ return properties[uint8(r)]&pLmask == pLl
+ }
+ return isExcludingLatin(Lower, r)
+}
+
+// IsTitle reports whether the rune is a title case letter.
+func IsTitle(r rune) bool {
+ if r <= MaxLatin1 {
+ return false
+ }
+ return isExcludingLatin(Title, r)
+}
+
+// to maps the rune using the specified case mapping.
+func to(_case int, r rune, caseRange []CaseRange) rune {
+ if _case < 0 || MaxCase <= _case {
+ return ReplacementChar // as reasonable an error as any
+ }
+ // binary search over ranges
+ lo := 0
+ hi := len(caseRange)
+ for lo < hi {
+ m := lo + (hi-lo)/2
+ cr := caseRange[m]
+ if rune(cr.Lo) <= r && r <= rune(cr.Hi) {
+ delta := rune(cr.Delta[_case])
+ if delta > MaxRune {
+ // In an Upper-Lower sequence, which always starts with
+ // an UpperCase letter, the real deltas always look like:
+ // {0, 1, 0} UpperCase (Lower is next)
+ // {-1, 0, -1} LowerCase (Upper, Title are previous)
+ // The characters at even offsets from the beginning of the
+ // sequence are upper case; the ones at odd offsets are lower.
+ // The correct mapping can be done by clearing or setting the low
+ // bit in the sequence offset.
+ // The constants UpperCase and TitleCase are even while LowerCase
+ // is odd so we take the low bit from _case.
+ return rune(cr.Lo) + ((r-rune(cr.Lo))&^1 | rune(_case&1))
+ }
+ return r + delta
+ }
+ if r < rune(cr.Lo) {
+ hi = m
+ } else {
+ lo = m + 1
+ }
+ }
+ return r
+}
+
+// To maps the rune to the specified case: UpperCase, LowerCase, or TitleCase.
+func To(_case int, r rune) rune {
+ return to(_case, r, CaseRanges)
+}
+
+// ToUpper maps the rune to upper case.
+func ToUpper(r rune) rune {
+ if r <= MaxASCII {
+ if 'a' <= r && r <= 'z' {
+ r -= 'a' - 'A'
+ }
+ return r
+ }
+ return To(UpperCase, r)
+}
+
+// ToLower maps the rune to lower case.
+func ToLower(r rune) rune {
+ if r <= MaxASCII {
+ if 'A' <= r && r <= 'Z' {
+ r += 'a' - 'A'
+ }
+ return r
+ }
+ return To(LowerCase, r)
+}
+
+// ToTitle maps the rune to title case.
+func ToTitle(r rune) rune {
+ if r <= MaxASCII {
+ if 'a' <= r && r <= 'z' { // title case is upper case for ASCII
+ r -= 'a' - 'A'
+ }
+ return r
+ }
+ return To(TitleCase, r)
+}
+
+// ToUpper maps the rune to upper case giving priority to the special mapping.
+func (special SpecialCase) ToUpper(r rune) rune {
+ r1 := to(UpperCase, r, []CaseRange(special))
+ if r1 == r {
+ r1 = ToUpper(r)
+ }
+ return r1
+}
+
+// ToTitle maps the rune to title case giving priority to the special mapping.
+func (special SpecialCase) ToTitle(r rune) rune {
+ r1 := to(TitleCase, r, []CaseRange(special))
+ if r1 == r {
+ r1 = ToTitle(r)
+ }
+ return r1
+}
+
+// ToLower maps the rune to lower case giving priority to the special mapping.
+func (special SpecialCase) ToLower(r rune) rune {
+ r1 := to(LowerCase, r, []CaseRange(special))
+ if r1 == r {
+ r1 = ToLower(r)
+ }
+ return r1
+}
+
+// caseOrbit is defined in tables.go as []foldPair. Right now all the
+// entries fit in uint16, so use uint16. If that changes, compilation
+// will fail (the constants in the composite literal will not fit in uint16)
+// and the types here can change to uint32.
+type foldPair struct {
+ From uint16
+ To uint16
+}
+
+// SimpleFold iterates over Unicode code points equivalent under
+// the Unicode-defined simple case folding. Among the code points
+// equivalent to rune (including rune itself), SimpleFold returns the
+// smallest rune > r if one exists, or else the smallest rune >= 0.
+//
+// For example:
+// SimpleFold('A') = 'a'
+// SimpleFold('a') = 'A'
+//
+// SimpleFold('K') = 'k'
+// SimpleFold('k') = '\u212A' (Kelvin symbol, K)
+// SimpleFold('\u212A') = 'K'
+//
+// SimpleFold('1') = '1'
+//
+func SimpleFold(r rune) rune {
+ // Consult caseOrbit table for special cases.
+ lo := 0
+ hi := len(caseOrbit)
+ for lo < hi {
+ m := lo + (hi-lo)/2
+ if rune(caseOrbit[m].From) < r {
+ lo = m + 1
+ } else {
+ hi = m
+ }
+ }
+ if lo < len(caseOrbit) && rune(caseOrbit[lo].From) == r {
+ return rune(caseOrbit[lo].To)
+ }
+
+ // No folding specified. This is a one- or two-element
+ // equivalence class containing rune and ToLower(rune)
+ // and ToUpper(rune) if they are different from rune.
+ if l := ToLower(r); l != r {
+ return l
+ }
+ return ToUpper(r)
+}
diff --git a/src/unicode/letter_test.go b/src/unicode/letter_test.go
new file mode 100644
index 000000000..4ee11fb36
--- /dev/null
+++ b/src/unicode/letter_test.go
@@ -0,0 +1,535 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode_test
+
+import (
+ "flag"
+ "fmt"
+ "runtime"
+ "sort"
+ "testing"
+ . "unicode"
+)
+
+var upperTest = []rune{
+ 0x41,
+ 0xc0,
+ 0xd8,
+ 0x100,
+ 0x139,
+ 0x14a,
+ 0x178,
+ 0x181,
+ 0x376,
+ 0x3cf,
+ 0x1f2a,
+ 0x2102,
+ 0x2c00,
+ 0x2c10,
+ 0x2c20,
+ 0xa650,
+ 0xa722,
+ 0xff3a,
+ 0x10400,
+ 0x1d400,
+ 0x1d7ca,
+}
+
+var notupperTest = []rune{
+ 0x40,
+ 0x5b,
+ 0x61,
+ 0x185,
+ 0x1b0,
+ 0x377,
+ 0x387,
+ 0x2150,
+ 0xffff,
+ 0x10000,
+}
+
+var letterTest = []rune{
+ 0x41,
+ 0x61,
+ 0xaa,
+ 0xba,
+ 0xc8,
+ 0xdb,
+ 0xf9,
+ 0x2ec,
+ 0x535,
+ 0x620,
+ 0x6e6,
+ 0x93d,
+ 0xa15,
+ 0xb99,
+ 0xdc0,
+ 0xedd,
+ 0x1000,
+ 0x1200,
+ 0x1312,
+ 0x1401,
+ 0x1885,
+ 0x2c00,
+ 0xa800,
+ 0xf900,
+ 0xfa30,
+ 0xffda,
+ 0xffdc,
+ 0x10000,
+ 0x10300,
+ 0x10400,
+ 0x20000,
+ 0x2f800,
+ 0x2fa1d,
+}
+
+var notletterTest = []rune{
+ 0x20,
+ 0x35,
+ 0x375,
+ 0x619,
+ 0x700,
+ 0xfffe,
+ 0x1ffff,
+ 0x10ffff,
+}
+
+// Contains all the special cased Latin-1 chars.
+var spaceTest = []rune{
+ 0x09,
+ 0x0a,
+ 0x0b,
+ 0x0c,
+ 0x0d,
+ 0x20,
+ 0x85,
+ 0xA0,
+ 0x2000,
+ 0x3000,
+}
+
+type caseT struct {
+ cas int
+ in, out rune
+}
+
+var caseTest = []caseT{
+ // errors
+ {-1, '\n', 0xFFFD},
+ {UpperCase, -1, -1},
+ {UpperCase, 1 << 30, 1 << 30},
+
+ // ASCII (special-cased so test carefully)
+ {UpperCase, '\n', '\n'},
+ {UpperCase, 'a', 'A'},
+ {UpperCase, 'A', 'A'},
+ {UpperCase, '7', '7'},
+ {LowerCase, '\n', '\n'},
+ {LowerCase, 'a', 'a'},
+ {LowerCase, 'A', 'a'},
+ {LowerCase, '7', '7'},
+ {TitleCase, '\n', '\n'},
+ {TitleCase, 'a', 'A'},
+ {TitleCase, 'A', 'A'},
+ {TitleCase, '7', '7'},
+
+ // Latin-1: easy to read the tests!
+ {UpperCase, 0x80, 0x80},
+ {UpperCase, 'Å', 'Å'},
+ {UpperCase, 'å', 'Å'},
+ {LowerCase, 0x80, 0x80},
+ {LowerCase, 'Å', 'å'},
+ {LowerCase, 'å', 'å'},
+ {TitleCase, 0x80, 0x80},
+ {TitleCase, 'Å', 'Å'},
+ {TitleCase, 'å', 'Å'},
+
+ // 0131;LATIN SMALL LETTER DOTLESS I;Ll;0;L;;;;;N;;;0049;;0049
+ {UpperCase, 0x0131, 'I'},
+ {LowerCase, 0x0131, 0x0131},
+ {TitleCase, 0x0131, 'I'},
+
+ // 0133;LATIN SMALL LIGATURE IJ;Ll;0;L;<compat> 0069 006A;;;;N;LATIN SMALL LETTER I J;;0132;;0132
+ {UpperCase, 0x0133, 0x0132},
+ {LowerCase, 0x0133, 0x0133},
+ {TitleCase, 0x0133, 0x0132},
+
+ // 212A;KELVIN SIGN;Lu;0;L;004B;;;;N;DEGREES KELVIN;;;006B;
+ {UpperCase, 0x212A, 0x212A},
+ {LowerCase, 0x212A, 'k'},
+ {TitleCase, 0x212A, 0x212A},
+
+ // From an UpperLower sequence
+ // A640;CYRILLIC CAPITAL LETTER ZEMLYA;Lu;0;L;;;;;N;;;;A641;
+ {UpperCase, 0xA640, 0xA640},
+ {LowerCase, 0xA640, 0xA641},
+ {TitleCase, 0xA640, 0xA640},
+ // A641;CYRILLIC SMALL LETTER ZEMLYA;Ll;0;L;;;;;N;;;A640;;A640
+ {UpperCase, 0xA641, 0xA640},
+ {LowerCase, 0xA641, 0xA641},
+ {TitleCase, 0xA641, 0xA640},
+ // A64E;CYRILLIC CAPITAL LETTER NEUTRAL YER;Lu;0;L;;;;;N;;;;A64F;
+ {UpperCase, 0xA64E, 0xA64E},
+ {LowerCase, 0xA64E, 0xA64F},
+ {TitleCase, 0xA64E, 0xA64E},
+ // A65F;CYRILLIC SMALL LETTER YN;Ll;0;L;;;;;N;;;A65E;;A65E
+ {UpperCase, 0xA65F, 0xA65E},
+ {LowerCase, 0xA65F, 0xA65F},
+ {TitleCase, 0xA65F, 0xA65E},
+
+ // From another UpperLower sequence
+ // 0139;LATIN CAPITAL LETTER L WITH ACUTE;Lu;0;L;004C 0301;;;;N;LATIN CAPITAL LETTER L ACUTE;;;013A;
+ {UpperCase, 0x0139, 0x0139},
+ {LowerCase, 0x0139, 0x013A},
+ {TitleCase, 0x0139, 0x0139},
+ // 013F;LATIN CAPITAL LETTER L WITH MIDDLE DOT;Lu;0;L;<compat> 004C 00B7;;;;N;;;;0140;
+ {UpperCase, 0x013f, 0x013f},
+ {LowerCase, 0x013f, 0x0140},
+ {TitleCase, 0x013f, 0x013f},
+ // 0148;LATIN SMALL LETTER N WITH CARON;Ll;0;L;006E 030C;;;;N;LATIN SMALL LETTER N HACEK;;0147;;0147
+ {UpperCase, 0x0148, 0x0147},
+ {LowerCase, 0x0148, 0x0148},
+ {TitleCase, 0x0148, 0x0147},
+
+ // Last block in the 5.1.0 table
+ // 10400;DESERET CAPITAL LETTER LONG I;Lu;0;L;;;;;N;;;;10428;
+ {UpperCase, 0x10400, 0x10400},
+ {LowerCase, 0x10400, 0x10428},
+ {TitleCase, 0x10400, 0x10400},
+ // 10427;DESERET CAPITAL LETTER EW;Lu;0;L;;;;;N;;;;1044F;
+ {UpperCase, 0x10427, 0x10427},
+ {LowerCase, 0x10427, 0x1044F},
+ {TitleCase, 0x10427, 0x10427},
+ // 10428;DESERET SMALL LETTER LONG I;Ll;0;L;;;;;N;;;10400;;10400
+ {UpperCase, 0x10428, 0x10400},
+ {LowerCase, 0x10428, 0x10428},
+ {TitleCase, 0x10428, 0x10400},
+ // 1044F;DESERET SMALL LETTER EW;Ll;0;L;;;;;N;;;10427;;10427
+ {UpperCase, 0x1044F, 0x10427},
+ {LowerCase, 0x1044F, 0x1044F},
+ {TitleCase, 0x1044F, 0x10427},
+
+ // First one not in the 5.1.0 table
+ // 10450;SHAVIAN LETTER PEEP;Lo;0;L;;;;;N;;;;;
+ {UpperCase, 0x10450, 0x10450},
+ {LowerCase, 0x10450, 0x10450},
+ {TitleCase, 0x10450, 0x10450},
+
+ // Non-letters with case.
+ {LowerCase, 0x2161, 0x2171},
+ {UpperCase, 0x0345, 0x0399},
+}
+
+func TestIsLetter(t *testing.T) {
+ for _, r := range upperTest {
+ if !IsLetter(r) {
+ t.Errorf("IsLetter(U+%04X) = false, want true", r)
+ }
+ }
+ for _, r := range letterTest {
+ if !IsLetter(r) {
+ t.Errorf("IsLetter(U+%04X) = false, want true", r)
+ }
+ }
+ for _, r := range notletterTest {
+ if IsLetter(r) {
+ t.Errorf("IsLetter(U+%04X) = true, want false", r)
+ }
+ }
+}
+
+func TestIsUpper(t *testing.T) {
+ for _, r := range upperTest {
+ if !IsUpper(r) {
+ t.Errorf("IsUpper(U+%04X) = false, want true", r)
+ }
+ }
+ for _, r := range notupperTest {
+ if IsUpper(r) {
+ t.Errorf("IsUpper(U+%04X) = true, want false", r)
+ }
+ }
+ for _, r := range notletterTest {
+ if IsUpper(r) {
+ t.Errorf("IsUpper(U+%04X) = true, want false", r)
+ }
+ }
+}
+
+func caseString(c int) string {
+ switch c {
+ case UpperCase:
+ return "UpperCase"
+ case LowerCase:
+ return "LowerCase"
+ case TitleCase:
+ return "TitleCase"
+ }
+ return "ErrorCase"
+}
+
+func TestTo(t *testing.T) {
+ for _, c := range caseTest {
+ r := To(c.cas, c.in)
+ if c.out != r {
+ t.Errorf("To(U+%04X, %s) = U+%04X want U+%04X", c.in, caseString(c.cas), r, c.out)
+ }
+ }
+}
+
+func TestToUpperCase(t *testing.T) {
+ for _, c := range caseTest {
+ if c.cas != UpperCase {
+ continue
+ }
+ r := ToUpper(c.in)
+ if c.out != r {
+ t.Errorf("ToUpper(U+%04X) = U+%04X want U+%04X", c.in, r, c.out)
+ }
+ }
+}
+
+func TestToLowerCase(t *testing.T) {
+ for _, c := range caseTest {
+ if c.cas != LowerCase {
+ continue
+ }
+ r := ToLower(c.in)
+ if c.out != r {
+ t.Errorf("ToLower(U+%04X) = U+%04X want U+%04X", c.in, r, c.out)
+ }
+ }
+}
+
+func TestToTitleCase(t *testing.T) {
+ for _, c := range caseTest {
+ if c.cas != TitleCase {
+ continue
+ }
+ r := ToTitle(c.in)
+ if c.out != r {
+ t.Errorf("ToTitle(U+%04X) = U+%04X want U+%04X", c.in, r, c.out)
+ }
+ }
+}
+
+func TestIsSpace(t *testing.T) {
+ for _, c := range spaceTest {
+ if !IsSpace(c) {
+ t.Errorf("IsSpace(U+%04X) = false; want true", c)
+ }
+ }
+ for _, c := range letterTest {
+ if IsSpace(c) {
+ t.Errorf("IsSpace(U+%04X) = true; want false", c)
+ }
+ }
+}
+
+// Check that the optimizations for IsLetter etc. agree with the tables.
+// We only need to check the Latin-1 range.
+func TestLetterOptimizations(t *testing.T) {
+ for i := rune(0); i <= MaxLatin1; i++ {
+ if Is(Letter, i) != IsLetter(i) {
+ t.Errorf("IsLetter(U+%04X) disagrees with Is(Letter)", i)
+ }
+ if Is(Upper, i) != IsUpper(i) {
+ t.Errorf("IsUpper(U+%04X) disagrees with Is(Upper)", i)
+ }
+ if Is(Lower, i) != IsLower(i) {
+ t.Errorf("IsLower(U+%04X) disagrees with Is(Lower)", i)
+ }
+ if Is(Title, i) != IsTitle(i) {
+ t.Errorf("IsTitle(U+%04X) disagrees with Is(Title)", i)
+ }
+ if Is(White_Space, i) != IsSpace(i) {
+ t.Errorf("IsSpace(U+%04X) disagrees with Is(White_Space)", i)
+ }
+ if To(UpperCase, i) != ToUpper(i) {
+ t.Errorf("ToUpper(U+%04X) disagrees with To(Upper)", i)
+ }
+ if To(LowerCase, i) != ToLower(i) {
+ t.Errorf("ToLower(U+%04X) disagrees with To(Lower)", i)
+ }
+ if To(TitleCase, i) != ToTitle(i) {
+ t.Errorf("ToTitle(U+%04X) disagrees with To(Title)", i)
+ }
+ }
+}
+
+func TestTurkishCase(t *testing.T) {
+ lower := []rune("abcçdefgğhıijklmnoöprsştuüvyz")
+ upper := []rune("ABCÇDEFGĞHIİJKLMNOÖPRSŞTUÜVYZ")
+ for i, l := range lower {
+ u := upper[i]
+ if TurkishCase.ToLower(l) != l {
+ t.Errorf("lower(U+%04X) is U+%04X not U+%04X", l, TurkishCase.ToLower(l), l)
+ }
+ if TurkishCase.ToUpper(u) != u {
+ t.Errorf("upper(U+%04X) is U+%04X not U+%04X", u, TurkishCase.ToUpper(u), u)
+ }
+ if TurkishCase.ToUpper(l) != u {
+ t.Errorf("upper(U+%04X) is U+%04X not U+%04X", l, TurkishCase.ToUpper(l), u)
+ }
+ if TurkishCase.ToLower(u) != l {
+ t.Errorf("lower(U+%04X) is U+%04X not U+%04X", u, TurkishCase.ToLower(l), l)
+ }
+ if TurkishCase.ToTitle(u) != u {
+ t.Errorf("title(U+%04X) is U+%04X not U+%04X", u, TurkishCase.ToTitle(u), u)
+ }
+ if TurkishCase.ToTitle(l) != u {
+ t.Errorf("title(U+%04X) is U+%04X not U+%04X", l, TurkishCase.ToTitle(l), u)
+ }
+ }
+}
+
+var simpleFoldTests = []string{
+ // SimpleFold(x) returns the next equivalent rune > x or wraps
+ // around to smaller values.
+
+ // Easy cases.
+ "Aa",
+ "δΔ",
+
+ // ASCII special cases.
+ "KkK",
+ "Ssſ",
+
+ // Non-ASCII special cases.
+ "ρϱΡ",
+ "ͅΙιι",
+
+ // Extra special cases: has lower/upper but no case fold.
+ "İ",
+ "ı",
+}
+
+func TestSimpleFold(t *testing.T) {
+ for _, tt := range simpleFoldTests {
+ cycle := []rune(tt)
+ r := cycle[len(cycle)-1]
+ for _, out := range cycle {
+ if r := SimpleFold(r); r != out {
+ t.Errorf("SimpleFold(%#U) = %#U, want %#U", r, r, out)
+ }
+ r = out
+ }
+ }
+}
+
+// Running 'go test -calibrate' runs the calibration to find a plausible
+// cutoff point for linear search of a range list vs. binary search.
+// We create a fake table and then time how long it takes to do a
+// sequence of searches within that table, for all possible inputs
+// relative to the ranges (something before all, in each, between each, after all).
+// This assumes that all possible runes are equally likely.
+// In practice most runes are ASCII so this is a conservative estimate
+// of an effective cutoff value. In practice we could probably set it higher
+// than what this function recommends.
+
+var calibrate = flag.Bool("calibrate", false, "compute crossover for linear vs. binary search")
+
+func TestCalibrate(t *testing.T) {
+ if !*calibrate {
+ return
+ }
+
+ if runtime.GOARCH == "amd64" {
+ fmt.Printf("warning: running calibration on %s\n", runtime.GOARCH)
+ }
+
+ // Find the point where binary search wins by more than 10%.
+ // The 10% bias gives linear search an edge when they're close,
+ // because on predominantly ASCII inputs linear search is even
+ // better than our benchmarks measure.
+ n := sort.Search(64, func(n int) bool {
+ tab := fakeTable(n)
+ blinear := func(b *testing.B) {
+ tab := tab
+ max := n*5 + 20
+ for i := 0; i < b.N; i++ {
+ for j := 0; j <= max; j++ {
+ linear(tab, uint16(j))
+ }
+ }
+ }
+ bbinary := func(b *testing.B) {
+ tab := tab
+ max := n*5 + 20
+ for i := 0; i < b.N; i++ {
+ for j := 0; j <= max; j++ {
+ binary(tab, uint16(j))
+ }
+ }
+ }
+ bmlinear := testing.Benchmark(blinear)
+ bmbinary := testing.Benchmark(bbinary)
+ fmt.Printf("n=%d: linear=%d binary=%d\n", n, bmlinear.NsPerOp(), bmbinary.NsPerOp())
+ return bmlinear.NsPerOp()*100 > bmbinary.NsPerOp()*110
+ })
+ fmt.Printf("calibration: linear cutoff = %d\n", n)
+}
+
+func fakeTable(n int) []Range16 {
+ var r16 []Range16
+ for i := 0; i < n; i++ {
+ r16 = append(r16, Range16{uint16(i*5 + 10), uint16(i*5 + 12), 1})
+ }
+ return r16
+}
+
+func linear(ranges []Range16, r uint16) bool {
+ for i := range ranges {
+ range_ := &ranges[i]
+ if r < range_.Lo {
+ return false
+ }
+ if r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ }
+ return false
+}
+
+func binary(ranges []Range16, r uint16) bool {
+ // binary search over ranges
+ lo := 0
+ hi := len(ranges)
+ for lo < hi {
+ m := lo + (hi-lo)/2
+ range_ := &ranges[m]
+ if range_.Lo <= r && r <= range_.Hi {
+ return (r-range_.Lo)%range_.Stride == 0
+ }
+ if r < range_.Lo {
+ hi = m
+ } else {
+ lo = m + 1
+ }
+ }
+ return false
+}
+
+func TestLatinOffset(t *testing.T) {
+ var maps = []map[string]*RangeTable{
+ Categories,
+ FoldCategory,
+ FoldScript,
+ Properties,
+ Scripts,
+ }
+ for _, m := range maps {
+ for name, tab := range m {
+ i := 0
+ for i < len(tab.R16) && tab.R16[i].Hi <= MaxLatin1 {
+ i++
+ }
+ if tab.LatinOffset != i {
+ t.Errorf("%s: LatinOffset=%d, want %d", name, tab.LatinOffset, i)
+ }
+ }
+ }
+}
diff --git a/src/unicode/maketables.go b/src/unicode/maketables.go
new file mode 100644
index 000000000..d1c9aa04a
--- /dev/null
+++ b/src/unicode/maketables.go
@@ -0,0 +1,1376 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build ignore
+
+// Unicode table generator.
+// Data read from the web.
+
+package main
+
+import (
+ "bufio"
+ "flag"
+ "fmt"
+ "io"
+ "log"
+ "net/http"
+ "os"
+ "os/exec"
+ "path/filepath"
+ "regexp"
+ "sort"
+ "strconv"
+ "strings"
+ "unicode"
+)
+
+func main() {
+ flag.Parse()
+ setupOutput()
+ loadChars() // always needed
+ loadCasefold()
+ printCategories()
+ printScriptOrProperty(false)
+ printScriptOrProperty(true)
+ printCases()
+ printLatinProperties()
+ printCasefold()
+ printSizes()
+ flushOutput()
+}
+
+var dataURL = flag.String("data", "", "full URL for UnicodeData.txt; defaults to --url/UnicodeData.txt")
+var casefoldingURL = flag.String("casefolding", "", "full URL for CaseFolding.txt; defaults to --url/CaseFolding.txt")
+var url = flag.String("url",
+ "http://www.unicode.org/Public/7.0.0/ucd/",
+ "URL of Unicode database directory")
+var tablelist = flag.String("tables",
+ "all",
+ "comma-separated list of which tables to generate; can be letter")
+var scriptlist = flag.String("scripts",
+ "all",
+ "comma-separated list of which script tables to generate")
+var proplist = flag.String("props",
+ "all",
+ "comma-separated list of which property tables to generate")
+var cases = flag.Bool("cases",
+ true,
+ "generate case tables")
+var test = flag.Bool("test",
+ false,
+ "test existing tables; can be used to compare web data with package data")
+var localFiles = flag.Bool("local",
+ false,
+ "data files have been copied to current directory; for debugging only")
+var outputFile = flag.String("output",
+ "",
+ "output file for generated tables; default stdout")
+
+var scriptRe = regexp.MustCompile(`^([0-9A-F]+)(\.\.[0-9A-F]+)? *; ([A-Za-z_]+)$`)
+var logger = log.New(os.Stderr, "", log.Lshortfile)
+
+var output *bufio.Writer // points to os.Stdout or to "gofmt > outputFile"
+
+func setupOutput() {
+ output = bufio.NewWriter(startGofmt())
+}
+
+// startGofmt connects output to a gofmt process if -output is set.
+func startGofmt() io.Writer {
+ if *outputFile == "" {
+ return os.Stdout
+ }
+ stdout, err := os.Create(*outputFile)
+ if err != nil {
+ logger.Fatal(err)
+ }
+ // Pipe output to gofmt.
+ gofmt := exec.Command("gofmt")
+ fd, err := gofmt.StdinPipe()
+ if err != nil {
+ logger.Fatal(err)
+ }
+ gofmt.Stdout = stdout
+ gofmt.Stderr = os.Stderr
+ err = gofmt.Start()
+ if err != nil {
+ logger.Fatal(err)
+ }
+ return fd
+}
+
+func flushOutput() {
+ err := output.Flush()
+ if err != nil {
+ logger.Fatal(err)
+ }
+}
+
+func printf(format string, args ...interface{}) {
+ fmt.Fprintf(output, format, args...)
+}
+
+func print(args ...interface{}) {
+ fmt.Fprint(output, args...)
+}
+
+func println(args ...interface{}) {
+ fmt.Fprintln(output, args...)
+}
+
+type reader struct {
+ *bufio.Reader
+ fd *os.File
+ resp *http.Response
+}
+
+func open(url string) *reader {
+ file := filepath.Base(url)
+ if *localFiles {
+ fd, err := os.Open(file)
+ if err != nil {
+ logger.Fatal(err)
+ }
+ return &reader{bufio.NewReader(fd), fd, nil}
+ }
+ resp, err := http.Get(url)
+ if err != nil {
+ logger.Fatal(err)
+ }
+ if resp.StatusCode != 200 {
+ logger.Fatalf("bad GET status for %s: %d", file, resp.Status)
+ }
+ return &reader{bufio.NewReader(resp.Body), nil, resp}
+
+}
+
+func (r *reader) close() {
+ if r.fd != nil {
+ r.fd.Close()
+ } else {
+ r.resp.Body.Close()
+ }
+}
+
+var category = map[string]bool{
+ // Nd Lu etc.
+ // We use one-character names to identify merged categories
+ "L": true, // Lu Ll Lt Lm Lo
+ "P": true, // Pc Pd Ps Pe Pu Pf Po
+ "M": true, // Mn Mc Me
+ "N": true, // Nd Nl No
+ "S": true, // Sm Sc Sk So
+ "Z": true, // Zs Zl Zp
+ "C": true, // Cc Cf Cs Co Cn
+}
+
+// UnicodeData.txt has form:
+// 0037;DIGIT SEVEN;Nd;0;EN;;7;7;7;N;;;;;
+// 007A;LATIN SMALL LETTER Z;Ll;0;L;;;;;N;;;005A;;005A
+// See http://www.unicode.org/reports/tr44/ for a full explanation
+// The fields:
+const (
+ FCodePoint = iota
+ FName
+ FGeneralCategory
+ FCanonicalCombiningClass
+ FBidiClass
+ FDecompositionTypeAndMapping
+ FNumericType
+ FNumericDigit // If a decimal digit.
+ FNumericValue // Includes non-decimal, e.g. U+2155=1/5
+ FBidiMirrored
+ FUnicode1Name
+ FISOComment
+ FSimpleUppercaseMapping
+ FSimpleLowercaseMapping
+ FSimpleTitlecaseMapping
+ NumField
+
+ MaxChar = 0x10FFFF // anything above this shouldn't exist
+)
+
+var fieldName = []string{
+ FCodePoint: "CodePoint",
+ FName: "Name",
+ FGeneralCategory: "GeneralCategory",
+ FCanonicalCombiningClass: "CanonicalCombiningClass",
+ FBidiClass: "BidiClass",
+ FDecompositionTypeAndMapping: "DecompositionTypeAndMapping",
+ FNumericType: "NumericType",
+ FNumericDigit: "NumericDigit",
+ FNumericValue: "NumericValue",
+ FBidiMirrored: "BidiMirrored",
+ FUnicode1Name: "Unicode1Name",
+ FISOComment: "ISOComment",
+ FSimpleUppercaseMapping: "SimpleUppercaseMapping",
+ FSimpleLowercaseMapping: "SimpleLowercaseMapping",
+ FSimpleTitlecaseMapping: "SimpleTitlecaseMapping",
+}
+
+// This contains only the properties we're interested in.
+type Char struct {
+ field []string // debugging only; could be deleted if we take out char.dump()
+ codePoint rune // if zero, this index is not a valid code point.
+ category string
+ upperCase rune
+ lowerCase rune
+ titleCase rune
+ foldCase rune // simple case folding
+ caseOrbit rune // next in simple case folding orbit
+}
+
+// Scripts.txt has form:
+// A673 ; Cyrillic # Po SLAVONIC ASTERISK
+// A67C..A67D ; Cyrillic # Mn [2] COMBINING CYRILLIC KAVYKA..COMBINING CYRILLIC PAYEROK
+// See http://www.unicode.org/Public/5.1.0/ucd/UCD.html for full explanation
+
+type Script struct {
+ lo, hi uint32 // range of code points
+ script string
+}
+
+var chars = make([]Char, MaxChar+1)
+var scripts = make(map[string][]Script)
+var props = make(map[string][]Script) // a property looks like a script; can share the format
+
+var lastChar rune = 0
+
+// In UnicodeData.txt, some ranges are marked like this:
+// 3400;<CJK Ideograph Extension A, First>;Lo;0;L;;;;;N;;;;;
+// 4DB5;<CJK Ideograph Extension A, Last>;Lo;0;L;;;;;N;;;;;
+// parseCategory returns a state variable indicating the weirdness.
+type State int
+
+const (
+ SNormal State = iota // known to be zero for the type
+ SFirst
+ SLast
+ SMissing
+)
+
+func parseCategory(line string) (state State) {
+ field := strings.Split(line, ";")
+ if len(field) != NumField {
+ logger.Fatalf("%5s: %d fields (expected %d)\n", line, len(field), NumField)
+ }
+ point, err := strconv.ParseUint(field[FCodePoint], 16, 64)
+ if err != nil {
+ logger.Fatalf("%.5s...: %s", line, err)
+ }
+ lastChar = rune(point)
+ if point == 0 {
+ return // not interesting and we use 0 as unset
+ }
+ if point > MaxChar {
+ return
+ }
+ char := &chars[point]
+ char.field = field
+ if char.codePoint != 0 {
+ logger.Fatalf("point %U reused", point)
+ }
+ char.codePoint = lastChar
+ char.category = field[FGeneralCategory]
+ category[char.category] = true
+ switch char.category {
+ case "Nd":
+ // Decimal digit
+ _, err := strconv.Atoi(field[FNumericValue])
+ if err != nil {
+ logger.Fatalf("%U: bad numeric field: %s", point, err)
+ }
+ case "Lu":
+ char.letter(field[FCodePoint], field[FSimpleLowercaseMapping], field[FSimpleTitlecaseMapping])
+ case "Ll":
+ char.letter(field[FSimpleUppercaseMapping], field[FCodePoint], field[FSimpleTitlecaseMapping])
+ case "Lt":
+ char.letter(field[FSimpleUppercaseMapping], field[FSimpleLowercaseMapping], field[FCodePoint])
+ default:
+ char.letter(field[FSimpleUppercaseMapping], field[FSimpleLowercaseMapping], field[FSimpleTitlecaseMapping])
+ }
+ switch {
+ case strings.Index(field[FName], ", First>") > 0:
+ state = SFirst
+ case strings.Index(field[FName], ", Last>") > 0:
+ state = SLast
+ }
+ return
+}
+
+func (char *Char) dump(s string) {
+ print(s, " ")
+ for i := 0; i < len(char.field); i++ {
+ printf("%s:%q ", fieldName[i], char.field[i])
+ }
+ print("\n")
+}
+
+func (char *Char) letter(u, l, t string) {
+ char.upperCase = char.letterValue(u, "U")
+ char.lowerCase = char.letterValue(l, "L")
+ char.titleCase = char.letterValue(t, "T")
+}
+
+func (char *Char) letterValue(s string, cas string) rune {
+ if s == "" {
+ return 0
+ }
+ v, err := strconv.ParseUint(s, 16, 64)
+ if err != nil {
+ char.dump(cas)
+ logger.Fatalf("%U: bad letter(%s): %s", char.codePoint, s, err)
+ }
+ return rune(v)
+}
+
+func allCategories() []string {
+ a := make([]string, 0, len(category))
+ for k := range category {
+ a = append(a, k)
+ }
+ sort.Strings(a)
+ return a
+}
+
+func all(scripts map[string][]Script) []string {
+ a := make([]string, 0, len(scripts))
+ for k := range scripts {
+ a = append(a, k)
+ }
+ sort.Strings(a)
+ return a
+}
+
+func allCatFold(m map[string]map[rune]bool) []string {
+ a := make([]string, 0, len(m))
+ for k := range m {
+ a = append(a, k)
+ }
+ sort.Strings(a)
+ return a
+}
+
+// Extract the version number from the URL
+func version() string {
+ // Break on slashes and look for the first numeric field
+ fields := strings.Split(*url, "/")
+ for _, f := range fields {
+ if len(f) > 0 && '0' <= f[0] && f[0] <= '9' {
+ return f
+ }
+ }
+ logger.Fatal("unknown version")
+ return "Unknown"
+}
+
+func categoryOp(code rune, class uint8) bool {
+ category := chars[code].category
+ return len(category) > 0 && category[0] == class
+}
+
+func loadChars() {
+ if *dataURL == "" {
+ flag.Set("data", *url+"UnicodeData.txt")
+ }
+ input := open(*dataURL)
+ defer input.close()
+ scanner := bufio.NewScanner(input)
+ var first rune = 0
+ for scanner.Scan() {
+ switch parseCategory(scanner.Text()) {
+ case SNormal:
+ if first != 0 {
+ logger.Fatalf("bad state normal at %U", lastChar)
+ }
+ case SFirst:
+ if first != 0 {
+ logger.Fatalf("bad state first at %U", lastChar)
+ }
+ first = lastChar
+ case SLast:
+ if first == 0 {
+ logger.Fatalf("bad state last at %U", lastChar)
+ }
+ for i := first + 1; i <= lastChar; i++ {
+ chars[i] = chars[first]
+ chars[i].codePoint = i
+ }
+ first = 0
+ }
+ }
+ if scanner.Err() != nil {
+ logger.Fatal(scanner.Err())
+ }
+}
+
+func loadCasefold() {
+ if *casefoldingURL == "" {
+ flag.Set("casefolding", *url+"CaseFolding.txt")
+ }
+ input := open(*casefoldingURL)
+ defer input.close()
+ scanner := bufio.NewScanner(input)
+ for scanner.Scan() {
+ line := scanner.Text()
+ if len(line) == 0 || line[0] == '#' || len(strings.TrimSpace(line)) == 0 {
+ continue
+ }
+ field := strings.Split(line, "; ")
+ if len(field) != 4 {
+ logger.Fatalf("CaseFolding.txt %.5s...: %d fields (expected %d)\n", line, len(field), 4)
+ }
+ kind := field[1]
+ if kind != "C" && kind != "S" {
+ // Only care about 'common' and 'simple' foldings.
+ continue
+ }
+ p1, err := strconv.ParseUint(field[0], 16, 64)
+ if err != nil {
+ logger.Fatalf("CaseFolding.txt %.5s...: %s", line, err)
+ }
+ p2, err := strconv.ParseUint(field[2], 16, 64)
+ if err != nil {
+ logger.Fatalf("CaseFolding.txt %.5s...: %s", line, err)
+ }
+ chars[p1].foldCase = rune(p2)
+ }
+ if scanner.Err() != nil {
+ logger.Fatal(scanner.Err())
+ }
+}
+
+const progHeader = `// Copyright 2013 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Generated by running
+// maketables --tables=%s --data=%s --casefolding=%s
+// DO NOT EDIT
+
+package unicode
+
+`
+
+func printCategories() {
+ if *tablelist == "" {
+ return
+ }
+ // Find out which categories to dump
+ list := strings.Split(*tablelist, ",")
+ if *tablelist == "all" {
+ list = allCategories()
+ }
+ if *test {
+ fullCategoryTest(list)
+ return
+ }
+ printf(progHeader, *tablelist, *dataURL, *casefoldingURL)
+
+ println("// Version is the Unicode edition from which the tables are derived.")
+ printf("const Version = %q\n\n", version())
+
+ if *tablelist == "all" {
+ println("// Categories is the set of Unicode category tables.")
+ println("var Categories = map[string] *RangeTable {")
+ for _, k := range allCategories() {
+ printf("\t%q: %s,\n", k, k)
+ }
+ print("}\n\n")
+ }
+
+ decl := make(sort.StringSlice, len(list))
+ ndecl := 0
+ for _, name := range list {
+ if _, ok := category[name]; !ok {
+ logger.Fatal("unknown category", name)
+ }
+ // We generate an UpperCase name to serve as concise documentation and an _UnderScored
+ // name to store the data. This stops godoc dumping all the tables but keeps them
+ // available to clients.
+ // Cases deserving special comments
+ varDecl := ""
+ switch name {
+ case "C":
+ varDecl = "\tOther = _C; // Other/C is the set of Unicode control and special characters, category C.\n"
+ varDecl += "\tC = _C\n"
+ case "L":
+ varDecl = "\tLetter = _L; // Letter/L is the set of Unicode letters, category L.\n"
+ varDecl += "\tL = _L\n"
+ case "M":
+ varDecl = "\tMark = _M; // Mark/M is the set of Unicode mark characters, category M.\n"
+ varDecl += "\tM = _M\n"
+ case "N":
+ varDecl = "\tNumber = _N; // Number/N is the set of Unicode number characters, category N.\n"
+ varDecl += "\tN = _N\n"
+ case "P":
+ varDecl = "\tPunct = _P; // Punct/P is the set of Unicode punctuation characters, category P.\n"
+ varDecl += "\tP = _P\n"
+ case "S":
+ varDecl = "\tSymbol = _S; // Symbol/S is the set of Unicode symbol characters, category S.\n"
+ varDecl += "\tS = _S\n"
+ case "Z":
+ varDecl = "\tSpace = _Z; // Space/Z is the set of Unicode space characters, category Z.\n"
+ varDecl += "\tZ = _Z\n"
+ case "Nd":
+ varDecl = "\tDigit = _Nd; // Digit is the set of Unicode characters with the \"decimal digit\" property.\n"
+ case "Lu":
+ varDecl = "\tUpper = _Lu; // Upper is the set of Unicode upper case letters.\n"
+ case "Ll":
+ varDecl = "\tLower = _Ll; // Lower is the set of Unicode lower case letters.\n"
+ case "Lt":
+ varDecl = "\tTitle = _Lt; // Title is the set of Unicode title case letters.\n"
+ }
+ if len(name) > 1 {
+ varDecl += fmt.Sprintf(
+ "\t%s = _%s; // %s is the set of Unicode characters in category %s.\n",
+ name, name, name, name)
+ }
+ decl[ndecl] = varDecl
+ ndecl++
+ if len(name) == 1 { // unified categories
+ decl := fmt.Sprintf("var _%s = &RangeTable{\n", name)
+ dumpRange(
+ decl,
+ func(code rune) bool { return categoryOp(code, name[0]) })
+ continue
+ }
+ dumpRange(
+ fmt.Sprintf("var _%s = &RangeTable{\n", name),
+ func(code rune) bool { return chars[code].category == name })
+ }
+ decl.Sort()
+ println("// These variables have type *RangeTable.")
+ println("var (")
+ for _, d := range decl {
+ print(d)
+ }
+ print(")\n\n")
+}
+
+type Op func(code rune) bool
+
+const format = "\t\t{0x%04x, 0x%04x, %d},\n"
+
+func dumpRange(header string, inCategory Op) {
+ print(header)
+ next := rune(0)
+ latinOffset := 0
+ print("\tR16: []Range16{\n")
+ // one Range for each iteration
+ count := &range16Count
+ size := 16
+ for {
+ // look for start of range
+ for next < rune(len(chars)) && !inCategory(next) {
+ next++
+ }
+ if next >= rune(len(chars)) {
+ // no characters remain
+ break
+ }
+
+ // start of range
+ lo := next
+ hi := next
+ stride := rune(1)
+ // accept lo
+ next++
+ // look for another character to set the stride
+ for next < rune(len(chars)) && !inCategory(next) {
+ next++
+ }
+ if next >= rune(len(chars)) {
+ // no more characters
+ printf(format, lo, hi, stride)
+ break
+ }
+ // set stride
+ stride = next - lo
+ // check for length of run. next points to first jump in stride
+ for i := next; i < rune(len(chars)); i++ {
+ if inCategory(i) == (((i - lo) % stride) == 0) {
+ // accept
+ if inCategory(i) {
+ hi = i
+ }
+ } else {
+ // no more characters in this run
+ break
+ }
+ }
+ if uint32(hi) <= unicode.MaxLatin1 {
+ latinOffset++
+ }
+ size, count = printRange(uint32(lo), uint32(hi), uint32(stride), size, count)
+ // next range: start looking where this range ends
+ next = hi + 1
+ }
+ print("\t},\n")
+ if latinOffset > 0 {
+ printf("\tLatinOffset: %d,\n", latinOffset)
+ }
+ print("}\n\n")
+}
+
+func printRange(lo, hi, stride uint32, size int, count *int) (int, *int) {
+ if size == 16 && hi >= 1<<16 {
+ if lo < 1<<16 {
+ if lo+stride != hi {
+ logger.Fatalf("unexpected straddle: %U %U %d", lo, hi, stride)
+ }
+ // No range contains U+FFFF as an instance, so split
+ // the range into two entries. That way we can maintain
+ // the invariant that R32 contains only >= 1<<16.
+ printf(format, lo, lo, 1)
+ lo = hi
+ stride = 1
+ *count++
+ }
+ print("\t},\n")
+ print("\tR32: []Range32{\n")
+ size = 32
+ count = &range32Count
+ }
+ printf(format, lo, hi, stride)
+ *count++
+ return size, count
+}
+
+func fullCategoryTest(list []string) {
+ for _, name := range list {
+ if _, ok := category[name]; !ok {
+ logger.Fatal("unknown category", name)
+ }
+ r, ok := unicode.Categories[name]
+ if !ok && len(name) > 1 {
+ logger.Fatalf("unknown table %q", name)
+ }
+ if len(name) == 1 {
+ verifyRange(name, func(code rune) bool { return categoryOp(code, name[0]) }, r)
+ } else {
+ verifyRange(
+ name,
+ func(code rune) bool { return chars[code].category == name },
+ r)
+ }
+ }
+}
+
+func verifyRange(name string, inCategory Op, table *unicode.RangeTable) {
+ count := 0
+ for j := range chars {
+ i := rune(j)
+ web := inCategory(i)
+ pkg := unicode.Is(table, i)
+ if web != pkg {
+ fmt.Fprintf(os.Stderr, "%s: %U: web=%t pkg=%t\n", name, i, web, pkg)
+ count++
+ if count > 10 {
+ break
+ }
+ }
+ }
+}
+
+func parseScript(line string, scripts map[string][]Script) {
+ comment := strings.Index(line, "#")
+ if comment >= 0 {
+ line = line[0:comment]
+ }
+ line = strings.TrimSpace(line)
+ if len(line) == 0 {
+ return
+ }
+ field := strings.Split(line, ";")
+ if len(field) != 2 {
+ logger.Fatalf("%s: %d fields (expected 2)\n", line, len(field))
+ }
+ matches := scriptRe.FindStringSubmatch(line)
+ if len(matches) != 4 {
+ logger.Fatalf("%s: %d matches (expected 3)\n", line, len(matches))
+ }
+ lo, err := strconv.ParseUint(matches[1], 16, 64)
+ if err != nil {
+ logger.Fatalf("%.5s...: %s", line, err)
+ }
+ hi := lo
+ if len(matches[2]) > 2 { // ignore leading ..
+ hi, err = strconv.ParseUint(matches[2][2:], 16, 64)
+ if err != nil {
+ logger.Fatalf("%.5s...: %s", line, err)
+ }
+ }
+ name := matches[3]
+ scripts[name] = append(scripts[name], Script{uint32(lo), uint32(hi), name})
+}
+
+// The script tables have a lot of adjacent elements. Fold them together.
+func foldAdjacent(r []Script) []unicode.Range32 {
+ s := make([]unicode.Range32, 0, len(r))
+ j := 0
+ for i := 0; i < len(r); i++ {
+ if j > 0 && r[i].lo == s[j-1].Hi+1 {
+ s[j-1].Hi = r[i].hi
+ } else {
+ s = s[0 : j+1]
+ s[j] = unicode.Range32{
+ Lo: uint32(r[i].lo),
+ Hi: uint32(r[i].hi),
+ Stride: 1,
+ }
+ j++
+ }
+ }
+ return s
+}
+
+func fullScriptTest(list []string, installed map[string]*unicode.RangeTable, scripts map[string][]Script) {
+ for _, name := range list {
+ if _, ok := scripts[name]; !ok {
+ logger.Fatal("unknown script", name)
+ }
+ _, ok := installed[name]
+ if !ok {
+ logger.Fatal("unknown table", name)
+ }
+ for _, script := range scripts[name] {
+ for r := script.lo; r <= script.hi; r++ {
+ if !unicode.Is(installed[name], rune(r)) {
+ fmt.Fprintf(os.Stderr, "%U: not in script %s\n", r, name)
+ }
+ }
+ }
+ }
+}
+
+// PropList.txt has the same format as Scripts.txt so we can share its parser.
+func printScriptOrProperty(doProps bool) {
+ flag := "scripts"
+ flaglist := *scriptlist
+ file := "Scripts.txt"
+ table := scripts
+ installed := unicode.Scripts
+ if doProps {
+ flag = "props"
+ flaglist = *proplist
+ file = "PropList.txt"
+ table = props
+ installed = unicode.Properties
+ }
+ if flaglist == "" {
+ return
+ }
+ input := open(*url + file)
+ scanner := bufio.NewScanner(input)
+ for scanner.Scan() {
+ parseScript(scanner.Text(), table)
+ }
+ if scanner.Err() != nil {
+ logger.Fatal(scanner.Err())
+ }
+ input.close()
+
+ // Find out which scripts to dump
+ list := strings.Split(flaglist, ",")
+ if flaglist == "all" {
+ list = all(table)
+ }
+ if *test {
+ fullScriptTest(list, installed, table)
+ return
+ }
+
+ printf(
+ "// Generated by running\n"+
+ "// maketables --%s=%s --url=%s\n"+
+ "// DO NOT EDIT\n\n",
+ flag,
+ flaglist,
+ *url)
+ if flaglist == "all" {
+ if doProps {
+ println("// Properties is the set of Unicode property tables.")
+ println("var Properties = map[string] *RangeTable{")
+ } else {
+ println("// Scripts is the set of Unicode script tables.")
+ println("var Scripts = map[string] *RangeTable{")
+ }
+ for _, k := range all(table) {
+ printf("\t%q: %s,\n", k, k)
+ }
+ print("}\n\n")
+ }
+
+ decl := make(sort.StringSlice, len(list))
+ ndecl := 0
+ for _, name := range list {
+ if doProps {
+ decl[ndecl] = fmt.Sprintf(
+ "\t%s = _%s;\t// %s is the set of Unicode characters with property %s.\n",
+ name, name, name, name)
+ } else {
+ decl[ndecl] = fmt.Sprintf(
+ "\t%s = _%s;\t// %s is the set of Unicode characters in script %s.\n",
+ name, name, name, name)
+ }
+ ndecl++
+ printf("var _%s = &RangeTable {\n", name)
+ ranges := foldAdjacent(table[name])
+ print("\tR16: []Range16{\n")
+ size := 16
+ count := &range16Count
+ for _, s := range ranges {
+ size, count = printRange(s.Lo, s.Hi, s.Stride, size, count)
+ }
+ print("\t},\n")
+ if off := findLatinOffset(ranges); off > 0 {
+ printf("\tLatinOffset: %d,\n", off)
+ }
+ print("}\n\n")
+ }
+ decl.Sort()
+ println("// These variables have type *RangeTable.")
+ println("var (")
+ for _, d := range decl {
+ print(d)
+ }
+ print(")\n\n")
+}
+
+func findLatinOffset(ranges []unicode.Range32) int {
+ i := 0
+ for i < len(ranges) && ranges[i].Hi <= unicode.MaxLatin1 {
+ i++
+ }
+ return i
+}
+
+const (
+ CaseUpper = 1 << iota
+ CaseLower
+ CaseTitle
+ CaseNone = 0 // must be zero
+ CaseMissing = -1 // character not present; not a valid case state
+)
+
+type caseState struct {
+ point rune
+ _case int
+ deltaToUpper rune
+ deltaToLower rune
+ deltaToTitle rune
+}
+
+// Is d a continuation of the state of c?
+func (c *caseState) adjacent(d *caseState) bool {
+ if d.point < c.point {
+ c, d = d, c
+ }
+ switch {
+ case d.point != c.point+1: // code points not adjacent (shouldn't happen)
+ return false
+ case d._case != c._case: // different cases
+ return c.upperLowerAdjacent(d)
+ case c._case == CaseNone:
+ return false
+ case c._case == CaseMissing:
+ return false
+ case d.deltaToUpper != c.deltaToUpper:
+ return false
+ case d.deltaToLower != c.deltaToLower:
+ return false
+ case d.deltaToTitle != c.deltaToTitle:
+ return false
+ }
+ return true
+}
+
+// Is d the same as c, but opposite in upper/lower case? this would make it
+// an element of an UpperLower sequence.
+func (c *caseState) upperLowerAdjacent(d *caseState) bool {
+ // check they're a matched case pair. we know they have adjacent values
+ switch {
+ case c._case == CaseUpper && d._case != CaseLower:
+ return false
+ case c._case == CaseLower && d._case != CaseUpper:
+ return false
+ }
+ // matched pair (at least in upper/lower). make the order Upper Lower
+ if c._case == CaseLower {
+ c, d = d, c
+ }
+ // for an Upper Lower sequence the deltas have to be in order
+ // c: 0 1 0
+ // d: -1 0 -1
+ switch {
+ case c.deltaToUpper != 0:
+ return false
+ case c.deltaToLower != 1:
+ return false
+ case c.deltaToTitle != 0:
+ return false
+ case d.deltaToUpper != -1:
+ return false
+ case d.deltaToLower != 0:
+ return false
+ case d.deltaToTitle != -1:
+ return false
+ }
+ return true
+}
+
+// Does this character start an UpperLower sequence?
+func (c *caseState) isUpperLower() bool {
+ // for an Upper Lower sequence the deltas have to be in order
+ // c: 0 1 0
+ switch {
+ case c.deltaToUpper != 0:
+ return false
+ case c.deltaToLower != 1:
+ return false
+ case c.deltaToTitle != 0:
+ return false
+ }
+ return true
+}
+
+// Does this character start a LowerUpper sequence?
+func (c *caseState) isLowerUpper() bool {
+ // for an Upper Lower sequence the deltas have to be in order
+ // c: -1 0 -1
+ switch {
+ case c.deltaToUpper != -1:
+ return false
+ case c.deltaToLower != 0:
+ return false
+ case c.deltaToTitle != -1:
+ return false
+ }
+ return true
+}
+
+func getCaseState(i rune) (c *caseState) {
+ c = &caseState{point: i, _case: CaseNone}
+ ch := &chars[i]
+ switch ch.codePoint {
+ case 0:
+ c._case = CaseMissing // Will get NUL wrong but that doesn't matter
+ return
+ case ch.upperCase:
+ c._case = CaseUpper
+ case ch.lowerCase:
+ c._case = CaseLower
+ case ch.titleCase:
+ c._case = CaseTitle
+ }
+ // Some things such as roman numeral U+2161 don't describe themselves
+ // as upper case, but have a lower case. Second-guess them.
+ if c._case == CaseNone && ch.lowerCase != 0 {
+ c._case = CaseUpper
+ }
+ // Same in the other direction.
+ if c._case == CaseNone && ch.upperCase != 0 {
+ c._case = CaseLower
+ }
+
+ if ch.upperCase != 0 {
+ c.deltaToUpper = ch.upperCase - i
+ }
+ if ch.lowerCase != 0 {
+ c.deltaToLower = ch.lowerCase - i
+ }
+ if ch.titleCase != 0 {
+ c.deltaToTitle = ch.titleCase - i
+ }
+ return
+}
+
+func printCases() {
+ if !*cases {
+ return
+ }
+ if *test {
+ fullCaseTest()
+ return
+ }
+ printf(
+ "// Generated by running\n"+
+ "// maketables --data=%s --casefolding=%s\n"+
+ "// DO NOT EDIT\n\n"+
+ "// CaseRanges is the table describing case mappings for all letters with\n"+
+ "// non-self mappings.\n"+
+ "var CaseRanges = _CaseRanges\n"+
+ "var _CaseRanges = []CaseRange {\n",
+ *dataURL, *casefoldingURL)
+
+ var startState *caseState // the start of a run; nil for not active
+ var prevState = &caseState{} // the state of the previous character
+ for i := range chars {
+ state := getCaseState(rune(i))
+ if state.adjacent(prevState) {
+ prevState = state
+ continue
+ }
+ // end of run (possibly)
+ printCaseRange(startState, prevState)
+ startState = nil
+ if state._case != CaseMissing && state._case != CaseNone {
+ startState = state
+ }
+ prevState = state
+ }
+ print("}\n")
+}
+
+func printCaseRange(lo, hi *caseState) {
+ if lo == nil {
+ return
+ }
+ if lo.deltaToUpper == 0 && lo.deltaToLower == 0 && lo.deltaToTitle == 0 {
+ // character represents itself in all cases - no need to mention it
+ return
+ }
+ switch {
+ case hi.point > lo.point && lo.isUpperLower():
+ printf("\t{0x%04X, 0x%04X, d{UpperLower, UpperLower, UpperLower}},\n",
+ lo.point, hi.point)
+ case hi.point > lo.point && lo.isLowerUpper():
+ logger.Fatalf("LowerUpper sequence: should not happen: %U. If it's real, need to fix To()", lo.point)
+ printf("\t{0x%04X, 0x%04X, d{LowerUpper, LowerUpper, LowerUpper}},\n",
+ lo.point, hi.point)
+ default:
+ printf("\t{0x%04X, 0x%04X, d{%d, %d, %d}},\n",
+ lo.point, hi.point,
+ lo.deltaToUpper, lo.deltaToLower, lo.deltaToTitle)
+ }
+}
+
+// If the cased value in the Char is 0, it means use the rune itself.
+func caseIt(r, cased rune) rune {
+ if cased == 0 {
+ return r
+ }
+ return cased
+}
+
+func fullCaseTest() {
+ for j, c := range chars {
+ i := rune(j)
+ lower := unicode.ToLower(i)
+ want := caseIt(i, c.lowerCase)
+ if lower != want {
+ fmt.Fprintf(os.Stderr, "lower %U should be %U is %U\n", i, want, lower)
+ }
+ upper := unicode.ToUpper(i)
+ want = caseIt(i, c.upperCase)
+ if upper != want {
+ fmt.Fprintf(os.Stderr, "upper %U should be %U is %U\n", i, want, upper)
+ }
+ title := unicode.ToTitle(i)
+ want = caseIt(i, c.titleCase)
+ if title != want {
+ fmt.Fprintf(os.Stderr, "title %U should be %U is %U\n", i, want, title)
+ }
+ }
+}
+
+func printLatinProperties() {
+ if *test {
+ return
+ }
+ println("var properties = [MaxLatin1+1]uint8{")
+ for code := 0; code <= unicode.MaxLatin1; code++ {
+ var property string
+ switch chars[code].category {
+ case "Cc", "": // NUL has no category.
+ property = "pC"
+ case "Cf": // soft hyphen, unique category, not printable.
+ property = "0"
+ case "Ll":
+ property = "pLl | pp"
+ case "Lo":
+ property = "pLo | pp"
+ case "Lu":
+ property = "pLu | pp"
+ case "Nd", "No":
+ property = "pN | pp"
+ case "Pc", "Pd", "Pe", "Pf", "Pi", "Po", "Ps":
+ property = "pP | pp"
+ case "Sc", "Sk", "Sm", "So":
+ property = "pS | pp"
+ case "Zs":
+ property = "pZ"
+ default:
+ logger.Fatalf("%U has unknown category %q", code, chars[code].category)
+ }
+ // Special case
+ if code == ' ' {
+ property = "pZ | pp"
+ }
+ printf("\t0x%02X: %s, // %q\n", code, property, code)
+ }
+ printf("}\n\n")
+}
+
+type runeSlice []rune
+
+func (p runeSlice) Len() int { return len(p) }
+func (p runeSlice) Less(i, j int) bool { return p[i] < p[j] }
+func (p runeSlice) Swap(i, j int) { p[i], p[j] = p[j], p[i] }
+
+func printCasefold() {
+ // Build list of case-folding groups attached to each canonical folded char (typically lower case).
+ var caseOrbit = make([][]rune, MaxChar+1)
+ for j := range chars {
+ i := rune(j)
+ c := &chars[i]
+ if c.foldCase == 0 {
+ continue
+ }
+ orb := caseOrbit[c.foldCase]
+ if orb == nil {
+ orb = append(orb, c.foldCase)
+ }
+ caseOrbit[c.foldCase] = append(orb, i)
+ }
+
+ // Insert explicit 1-element groups when assuming [lower, upper] would be wrong.
+ for j := range chars {
+ i := rune(j)
+ c := &chars[i]
+ f := c.foldCase
+ if f == 0 {
+ f = i
+ }
+ orb := caseOrbit[f]
+ if orb == nil && (c.upperCase != 0 && c.upperCase != i || c.lowerCase != 0 && c.lowerCase != i) {
+ // Default assumption of [upper, lower] is wrong.
+ caseOrbit[i] = []rune{i}
+ }
+ }
+
+ // Delete the groups for which assuming [lower, upper] is right.
+ for i, orb := range caseOrbit {
+ if len(orb) == 2 && chars[orb[0]].upperCase == orb[1] && chars[orb[1]].lowerCase == orb[0] {
+ caseOrbit[i] = nil
+ }
+ }
+
+ // Record orbit information in chars.
+ for _, orb := range caseOrbit {
+ if orb == nil {
+ continue
+ }
+ sort.Sort(runeSlice(orb))
+ c := orb[len(orb)-1]
+ for _, d := range orb {
+ chars[c].caseOrbit = d
+ c = d
+ }
+ }
+
+ printCaseOrbit()
+
+ // Tables of category and script folding exceptions: code points
+ // that must be added when interpreting a particular category/script
+ // in a case-folding context.
+ cat := make(map[string]map[rune]bool)
+ for name := range category {
+ if x := foldExceptions(inCategory(name)); len(x) > 0 {
+ cat[name] = x
+ }
+ }
+
+ scr := make(map[string]map[rune]bool)
+ for name := range scripts {
+ if x := foldExceptions(inScript(name)); len(x) > 0 {
+ cat[name] = x
+ }
+ }
+
+ printCatFold("FoldCategory", cat)
+ printCatFold("FoldScript", scr)
+}
+
+// inCategory returns a list of all the runes in the category.
+func inCategory(name string) []rune {
+ var x []rune
+ for j := range chars {
+ i := rune(j)
+ c := &chars[i]
+ if c.category == name || len(name) == 1 && len(c.category) > 1 && c.category[0] == name[0] {
+ x = append(x, i)
+ }
+ }
+ return x
+}
+
+// inScript returns a list of all the runes in the script.
+func inScript(name string) []rune {
+ var x []rune
+ for _, s := range scripts[name] {
+ for c := s.lo; c <= s.hi; c++ {
+ x = append(x, rune(c))
+ }
+ }
+ return x
+}
+
+// foldExceptions returns a list of all the runes fold-equivalent
+// to runes in class but not in class themselves.
+func foldExceptions(class []rune) map[rune]bool {
+ // Create map containing class and all fold-equivalent chars.
+ m := make(map[rune]bool)
+ for _, r := range class {
+ c := &chars[r]
+ if c.caseOrbit == 0 {
+ // Just upper and lower.
+ if u := c.upperCase; u != 0 {
+ m[u] = true
+ }
+ if l := c.lowerCase; l != 0 {
+ m[l] = true
+ }
+ m[r] = true
+ continue
+ }
+ // Otherwise walk orbit.
+ r0 := r
+ for {
+ m[r] = true
+ r = chars[r].caseOrbit
+ if r == r0 {
+ break
+ }
+ }
+ }
+
+ // Remove class itself.
+ for _, r := range class {
+ delete(m, r)
+ }
+
+ // What's left is the exceptions.
+ return m
+}
+
+var comment = map[string]string{
+ "FoldCategory": "// FoldCategory maps a category name to a table of\n" +
+ "// code points outside the category that are equivalent under\n" +
+ "// simple case folding to code points inside the category.\n" +
+ "// If there is no entry for a category name, there are no such points.\n",
+
+ "FoldScript": "// FoldScript maps a script name to a table of\n" +
+ "// code points outside the script that are equivalent under\n" +
+ "// simple case folding to code points inside the script.\n" +
+ "// If there is no entry for a script name, there are no such points.\n",
+}
+
+func printCaseOrbit() {
+ if *test {
+ for j := range chars {
+ i := rune(j)
+ c := &chars[i]
+ f := c.caseOrbit
+ if f == 0 {
+ if c.lowerCase != i && c.lowerCase != 0 {
+ f = c.lowerCase
+ } else if c.upperCase != i && c.upperCase != 0 {
+ f = c.upperCase
+ } else {
+ f = i
+ }
+ }
+ if g := unicode.SimpleFold(i); g != f {
+ fmt.Fprintf(os.Stderr, "unicode.SimpleFold(%#U) = %#U, want %#U\n", i, g, f)
+ }
+ }
+ return
+ }
+
+ printf("var caseOrbit = []foldPair{\n")
+ for i := range chars {
+ c := &chars[i]
+ if c.caseOrbit != 0 {
+ printf("\t{0x%04X, 0x%04X},\n", i, c.caseOrbit)
+ foldPairCount++
+ }
+ }
+ printf("}\n\n")
+}
+
+func printCatFold(name string, m map[string]map[rune]bool) {
+ if *test {
+ var pkgMap map[string]*unicode.RangeTable
+ if name == "FoldCategory" {
+ pkgMap = unicode.FoldCategory
+ } else {
+ pkgMap = unicode.FoldScript
+ }
+ if len(pkgMap) != len(m) {
+ fmt.Fprintf(os.Stderr, "unicode.%s has %d elements, want %d\n", name, len(pkgMap), len(m))
+ return
+ }
+ for k, v := range m {
+ t, ok := pkgMap[k]
+ if !ok {
+ fmt.Fprintf(os.Stderr, "unicode.%s[%q] missing\n", name, k)
+ continue
+ }
+ n := 0
+ for _, r := range t.R16 {
+ for c := rune(r.Lo); c <= rune(r.Hi); c += rune(r.Stride) {
+ if !v[c] {
+ fmt.Fprintf(os.Stderr, "unicode.%s[%q] contains %#U, should not\n", name, k, c)
+ }
+ n++
+ }
+ }
+ for _, r := range t.R32 {
+ for c := rune(r.Lo); c <= rune(r.Hi); c += rune(r.Stride) {
+ if !v[c] {
+ fmt.Fprintf(os.Stderr, "unicode.%s[%q] contains %#U, should not\n", name, k, c)
+ }
+ n++
+ }
+ }
+ if n != len(v) {
+ fmt.Fprintf(os.Stderr, "unicode.%s[%q] has %d code points, want %d\n", name, k, n, len(v))
+ }
+ }
+ return
+ }
+
+ print(comment[name])
+ printf("var %s = map[string]*RangeTable{\n", name)
+ for _, name := range allCatFold(m) {
+ printf("\t%q: fold%s,\n", name, name)
+ }
+ printf("}\n\n")
+ for _, name := range allCatFold(m) {
+ class := m[name]
+ dumpRange(
+ fmt.Sprintf("var fold%s = &RangeTable{\n", name),
+ func(code rune) bool { return class[code] })
+ }
+}
+
+var range16Count = 0 // Number of entries in the 16-bit range tables.
+var range32Count = 0 // Number of entries in the 32-bit range tables.
+var foldPairCount = 0 // Number of fold pairs in the exception tables.
+
+func printSizes() {
+ if *test {
+ return
+ }
+ println()
+ printf("// Range entries: %d 16-bit, %d 32-bit, %d total.\n", range16Count, range32Count, range16Count+range32Count)
+ range16Bytes := range16Count * 3 * 2
+ range32Bytes := range32Count * 3 * 4
+ printf("// Range bytes: %d 16-bit, %d 32-bit, %d total.\n", range16Bytes, range32Bytes, range16Bytes+range32Bytes)
+ println()
+ printf("// Fold orbit bytes: %d pairs, %d bytes\n", foldPairCount, foldPairCount*2*2)
+}
diff --git a/src/unicode/script_test.go b/src/unicode/script_test.go
new file mode 100644
index 000000000..795cb4e17
--- /dev/null
+++ b/src/unicode/script_test.go
@@ -0,0 +1,287 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package unicode_test
+
+import (
+ "testing"
+ . "unicode"
+)
+
+type T struct {
+ rune rune
+ script string
+}
+
+// Hand-chosen tests from Unicode 5.1.0, 6.0.0, 6.2.0, 6.3.0 and 7.0.0 mostly to
+// discover when new scripts and categories arise.
+var inTest = []T{
+ {0x06e2, "Arabic"},
+ {0x0567, "Armenian"},
+ {0x10b20, "Avestan"},
+ {0x1b37, "Balinese"},
+ {0xa6af, "Bamum"},
+ {0x16ada, "Bassa_Vah"},
+ {0x1be1, "Batak"},
+ {0x09c2, "Bengali"},
+ {0x3115, "Bopomofo"},
+ {0x282d, "Braille"},
+ {0x1a1a, "Buginese"},
+ {0x1747, "Buhid"},
+ {0x11011, "Brahmi"},
+ {0x156d, "Canadian_Aboriginal"},
+ {0x102a9, "Carian"},
+ {0x10563, "Caucasian_Albanian"},
+ {0x11111, "Chakma"},
+ {0xaa4d, "Cham"},
+ {0x13c2, "Cherokee"},
+ {0x0020, "Common"},
+ {0x1d4a5, "Common"},
+ {0x2cfc, "Coptic"},
+ {0x12420, "Cuneiform"},
+ {0x1080c, "Cypriot"},
+ {0xa663, "Cyrillic"},
+ {0x10430, "Deseret"},
+ {0x094a, "Devanagari"},
+ {0x1BC00, "Duployan"},
+ {0x13001, "Egyptian_Hieroglyphs"},
+ {0x10500, "Elbasan"},
+ {0x1271, "Ethiopic"},
+ {0x10fc, "Georgian"},
+ {0x2c40, "Glagolitic"},
+ {0x10347, "Gothic"},
+ {0x11303, "Grantha"},
+ {0x03ae, "Greek"},
+ {0x0abf, "Gujarati"},
+ {0x0a24, "Gurmukhi"},
+ {0x3028, "Han"},
+ {0x11b8, "Hangul"},
+ {0x1727, "Hanunoo"},
+ {0x05a0, "Hebrew"},
+ {0x3058, "Hiragana"},
+ {0x10841, "Imperial_Aramaic"},
+ {0x20e6, "Inherited"},
+ {0x10b70, "Inscriptional_Pahlavi"},
+ {0x10b5a, "Inscriptional_Parthian"},
+ {0xa9d0, "Javanese"},
+ {0x1109f, "Kaithi"},
+ {0x0cbd, "Kannada"},
+ {0x30a6, "Katakana"},
+ {0xa928, "Kayah_Li"},
+ {0x10a11, "Kharoshthi"},
+ {0x17c6, "Khmer"},
+ {0x11211, "Khojki"},
+ {0x112df, "Khudawadi"},
+ {0x0eaa, "Lao"},
+ {0x1d79, "Latin"},
+ {0x1c10, "Lepcha"},
+ {0x1930, "Limbu"},
+ {0x10755, "Linear_A"},
+ {0x1003c, "Linear_B"},
+ {0xa4e1, "Lisu"},
+ {0x10290, "Lycian"},
+ {0x10930, "Lydian"},
+ {0x11173, "Mahajani"},
+ {0x0d42, "Malayalam"},
+ {0x0843, "Mandaic"},
+ {0x10ac8, "Manichaean"},
+ {0xabd0, "Meetei_Mayek"},
+ {0x1e800, "Mende_Kikakui"},
+ {0x1099f, "Meroitic_Hieroglyphs"},
+ {0x109a0, "Meroitic_Cursive"},
+ {0x16f00, "Miao"},
+ {0x11611, "Modi"},
+ {0x1822, "Mongolian"},
+ {0x16a60, "Mro"},
+ {0x104c, "Myanmar"},
+ {0x10880, "Nabataean"},
+ {0x19c3, "New_Tai_Lue"},
+ {0x07f8, "Nko"},
+ {0x169b, "Ogham"},
+ {0x1c6a, "Ol_Chiki"},
+ {0x10310, "Old_Italic"},
+ {0x10a80, "Old_North_Arabian"},
+ {0x10350, "Old_Permic"},
+ {0x103c9, "Old_Persian"},
+ {0x10a6f, "Old_South_Arabian"},
+ {0x10c20, "Old_Turkic"},
+ {0x0b3e, "Oriya"},
+ {0x10491, "Osmanya"},
+ {0x16b2b, "Pahawh_Hmong"},
+ {0x10876, "Palmyrene"},
+ {0x11ACE, "Pau_Cin_Hau"},
+ {0xa860, "Phags_Pa"},
+ {0x10918, "Phoenician"},
+ {0x10baf, "Psalter_Pahlavi"},
+ {0xa949, "Rejang"},
+ {0x16c0, "Runic"},
+ {0x081d, "Samaritan"},
+ {0xa892, "Saurashtra"},
+ {0x111a0, "Sharada"},
+ {0x10463, "Shavian"},
+ {0x115c1, "Siddham"},
+ {0x0dbd, "Sinhala"},
+ {0x110d0, "Sora_Sompeng"},
+ {0x1ba3, "Sundanese"},
+ {0xa803, "Syloti_Nagri"},
+ {0x070f, "Syriac"},
+ {0x170f, "Tagalog"},
+ {0x176f, "Tagbanwa"},
+ {0x1972, "Tai_Le"},
+ {0x1a62, "Tai_Tham"},
+ {0xaadc, "Tai_Viet"},
+ {0x116c9, "Takri"},
+ {0x0bbf, "Tamil"},
+ {0x0c55, "Telugu"},
+ {0x07a7, "Thaana"},
+ {0x0e46, "Thai"},
+ {0x0f36, "Tibetan"},
+ {0x2d55, "Tifinagh"},
+ {0x114d9, "Tirhuta"},
+ {0x10388, "Ugaritic"},
+ {0xa60e, "Vai"},
+ {0x118ff, "Warang_Citi"},
+ {0xa216, "Yi"},
+}
+
+var outTest = []T{ // not really worth being thorough
+ {0x20, "Telugu"},
+}
+
+var inCategoryTest = []T{
+ {0x0081, "Cc"},
+ {0x200B, "Cf"},
+ {0xf0000, "Co"},
+ {0xdb80, "Cs"},
+ {0x0236, "Ll"},
+ {0x1d9d, "Lm"},
+ {0x07cf, "Lo"},
+ {0x1f8a, "Lt"},
+ {0x03ff, "Lu"},
+ {0x0bc1, "Mc"},
+ {0x20df, "Me"},
+ {0x07f0, "Mn"},
+ {0x1bb2, "Nd"},
+ {0x10147, "Nl"},
+ {0x2478, "No"},
+ {0xfe33, "Pc"},
+ {0x2011, "Pd"},
+ {0x301e, "Pe"},
+ {0x2e03, "Pf"},
+ {0x2e02, "Pi"},
+ {0x0022, "Po"},
+ {0x2770, "Ps"},
+ {0x00a4, "Sc"},
+ {0xa711, "Sk"},
+ {0x25f9, "Sm"},
+ {0x2108, "So"},
+ {0x2028, "Zl"},
+ {0x2029, "Zp"},
+ {0x202f, "Zs"},
+ // Unifieds.
+ {0x04aa, "L"},
+ {0x0009, "C"},
+ {0x1712, "M"},
+ {0x0031, "N"},
+ {0x00bb, "P"},
+ {0x00a2, "S"},
+ {0x00a0, "Z"},
+}
+
+var inPropTest = []T{
+ {0x0046, "ASCII_Hex_Digit"},
+ {0x200F, "Bidi_Control"},
+ {0x2212, "Dash"},
+ {0xE0001, "Deprecated"},
+ {0x00B7, "Diacritic"},
+ {0x30FE, "Extender"},
+ {0xFF46, "Hex_Digit"},
+ {0x2E17, "Hyphen"},
+ {0x2FFB, "IDS_Binary_Operator"},
+ {0x2FF3, "IDS_Trinary_Operator"},
+ {0xFA6A, "Ideographic"},
+ {0x200D, "Join_Control"},
+ {0x0EC4, "Logical_Order_Exception"},
+ {0x2FFFF, "Noncharacter_Code_Point"},
+ {0x065E, "Other_Alphabetic"},
+ {0x2065, "Other_Default_Ignorable_Code_Point"},
+ {0x0BD7, "Other_Grapheme_Extend"},
+ {0x0387, "Other_ID_Continue"},
+ {0x212E, "Other_ID_Start"},
+ {0x2094, "Other_Lowercase"},
+ {0x2040, "Other_Math"},
+ {0x216F, "Other_Uppercase"},
+ {0x0027, "Pattern_Syntax"},
+ {0x0020, "Pattern_White_Space"},
+ {0x300D, "Quotation_Mark"},
+ {0x2EF3, "Radical"},
+ {0x061F, "STerm"},
+ {0x2071, "Soft_Dotted"},
+ {0x003A, "Terminal_Punctuation"},
+ {0x9FC3, "Unified_Ideograph"},
+ {0xFE0F, "Variation_Selector"},
+ {0x0020, "White_Space"},
+}
+
+func TestScripts(t *testing.T) {
+ notTested := make(map[string]bool)
+ for k := range Scripts {
+ notTested[k] = true
+ }
+ for _, test := range inTest {
+ if _, ok := Scripts[test.script]; !ok {
+ t.Fatal(test.script, "not a known script")
+ }
+ if !Is(Scripts[test.script], test.rune) {
+ t.Errorf("IsScript(%U, %s) = false, want true", test.rune, test.script)
+ }
+ delete(notTested, test.script)
+ }
+ for _, test := range outTest {
+ if Is(Scripts[test.script], test.rune) {
+ t.Errorf("IsScript(%U, %s) = true, want false", test.rune, test.script)
+ }
+ }
+ for k := range notTested {
+ t.Error("script not tested:", k)
+ }
+}
+
+func TestCategories(t *testing.T) {
+ notTested := make(map[string]bool)
+ for k := range Categories {
+ notTested[k] = true
+ }
+ for _, test := range inCategoryTest {
+ if _, ok := Categories[test.script]; !ok {
+ t.Fatal(test.script, "not a known category")
+ }
+ if !Is(Categories[test.script], test.rune) {
+ t.Errorf("IsCategory(%U, %s) = false, want true", test.rune, test.script)
+ }
+ delete(notTested, test.script)
+ }
+ for k := range notTested {
+ t.Error("category not tested:", k)
+ }
+}
+
+func TestProperties(t *testing.T) {
+ notTested := make(map[string]bool)
+ for k := range Properties {
+ notTested[k] = true
+ }
+ for _, test := range inPropTest {
+ if _, ok := Properties[test.script]; !ok {
+ t.Fatal(test.script, "not a known prop")
+ }
+ if !Is(Properties[test.script], test.rune) {
+ t.Errorf("IsCategory(%U, %s) = false, want true", test.rune, test.script)
+ }
+ delete(notTested, test.script)
+ }
+ for k := range notTested {
+ t.Error("property not tested:", k)
+ }
+}
diff --git a/src/unicode/tables.go b/src/unicode/tables.go
new file mode 100644
index 000000000..8b77dd603
--- /dev/null
+++ b/src/unicode/tables.go
@@ -0,0 +1,7029 @@
+// Copyright 2013 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Generated by running
+// maketables --tables=all --data=http://www.unicode.org/Public/7.0.0/ucd/UnicodeData.txt --casefolding=http://www.unicode.org/Public/7.0.0/ucd/CaseFolding.txt
+// DO NOT EDIT
+
+package unicode
+
+// Version is the Unicode edition from which the tables are derived.
+const Version = "7.0.0"
+
+// Categories is the set of Unicode category tables.
+var Categories = map[string]*RangeTable{
+ "C": C,
+ "Cc": Cc,
+ "Cf": Cf,
+ "Co": Co,
+ "Cs": Cs,
+ "L": L,
+ "Ll": Ll,
+ "Lm": Lm,
+ "Lo": Lo,
+ "Lt": Lt,
+ "Lu": Lu,
+ "M": M,
+ "Mc": Mc,
+ "Me": Me,
+ "Mn": Mn,
+ "N": N,
+ "Nd": Nd,
+ "Nl": Nl,
+ "No": No,
+ "P": P,
+ "Pc": Pc,
+ "Pd": Pd,
+ "Pe": Pe,
+ "Pf": Pf,
+ "Pi": Pi,
+ "Po": Po,
+ "Ps": Ps,
+ "S": S,
+ "Sc": Sc,
+ "Sk": Sk,
+ "Sm": Sm,
+ "So": So,
+ "Z": Z,
+ "Zl": Zl,
+ "Zp": Zp,
+ "Zs": Zs,
+}
+
+var _C = &RangeTable{
+ R16: []Range16{
+ {0x0001, 0x001f, 1},
+ {0x007f, 0x009f, 1},
+ {0x00ad, 0x0600, 1363},
+ {0x0601, 0x0605, 1},
+ {0x061c, 0x06dd, 193},
+ {0x070f, 0x180e, 4351},
+ {0x200b, 0x200f, 1},
+ {0x202a, 0x202e, 1},
+ {0x2060, 0x2064, 1},
+ {0x2066, 0x206f, 1},
+ {0xd800, 0xf8ff, 1},
+ {0xfeff, 0xfff9, 250},
+ {0xfffa, 0xfffb, 1},
+ },
+ R32: []Range32{
+ {0x110bd, 0x1bca0, 44003},
+ {0x1bca1, 0x1bca3, 1},
+ {0x1d173, 0x1d17a, 1},
+ {0xe0001, 0xe0020, 31},
+ {0xe0021, 0xe007f, 1},
+ {0xf0000, 0xffffd, 1},
+ {0x100000, 0x10fffd, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _Cc = &RangeTable{
+ R16: []Range16{
+ {0x0001, 0x001f, 1},
+ {0x007f, 0x009f, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _Cf = &RangeTable{
+ R16: []Range16{
+ {0x00ad, 0x0600, 1363},
+ {0x0601, 0x0605, 1},
+ {0x061c, 0x06dd, 193},
+ {0x070f, 0x180e, 4351},
+ {0x200b, 0x200f, 1},
+ {0x202a, 0x202e, 1},
+ {0x2060, 0x2064, 1},
+ {0x2066, 0x206f, 1},
+ {0xfeff, 0xfff9, 250},
+ {0xfffa, 0xfffb, 1},
+ },
+ R32: []Range32{
+ {0x110bd, 0x1bca0, 44003},
+ {0x1bca1, 0x1bca3, 1},
+ {0x1d173, 0x1d17a, 1},
+ {0xe0001, 0xe0020, 31},
+ {0xe0021, 0xe007f, 1},
+ },
+}
+
+var _Co = &RangeTable{
+ R16: []Range16{
+ {0xe000, 0xf8ff, 1},
+ },
+ R32: []Range32{
+ {0xf0000, 0xffffd, 1},
+ {0x100000, 0x10fffd, 1},
+ },
+}
+
+var _Cs = &RangeTable{
+ R16: []Range16{
+ {0xd800, 0xdfff, 1},
+ },
+}
+
+var _L = &RangeTable{
+ R16: []Range16{
+ {0x0041, 0x005a, 1},
+ {0x0061, 0x007a, 1},
+ {0x00aa, 0x00b5, 11},
+ {0x00ba, 0x00c0, 6},
+ {0x00c1, 0x00d6, 1},
+ {0x00d8, 0x00f6, 1},
+ {0x00f8, 0x02c1, 1},
+ {0x02c6, 0x02d1, 1},
+ {0x02e0, 0x02e4, 1},
+ {0x02ec, 0x02ee, 2},
+ {0x0370, 0x0374, 1},
+ {0x0376, 0x0377, 1},
+ {0x037a, 0x037d, 1},
+ {0x037f, 0x0386, 7},
+ {0x0388, 0x038a, 1},
+ {0x038c, 0x038e, 2},
+ {0x038f, 0x03a1, 1},
+ {0x03a3, 0x03f5, 1},
+ {0x03f7, 0x0481, 1},
+ {0x048a, 0x052f, 1},
+ {0x0531, 0x0556, 1},
+ {0x0559, 0x0561, 8},
+ {0x0562, 0x0587, 1},
+ {0x05d0, 0x05ea, 1},
+ {0x05f0, 0x05f2, 1},
+ {0x0620, 0x064a, 1},
+ {0x066e, 0x066f, 1},
+ {0x0671, 0x06d3, 1},
+ {0x06d5, 0x06e5, 16},
+ {0x06e6, 0x06ee, 8},
+ {0x06ef, 0x06fa, 11},
+ {0x06fb, 0x06fc, 1},
+ {0x06ff, 0x0710, 17},
+ {0x0712, 0x072f, 1},
+ {0x074d, 0x07a5, 1},
+ {0x07b1, 0x07ca, 25},
+ {0x07cb, 0x07ea, 1},
+ {0x07f4, 0x07f5, 1},
+ {0x07fa, 0x0800, 6},
+ {0x0801, 0x0815, 1},
+ {0x081a, 0x0824, 10},
+ {0x0828, 0x0840, 24},
+ {0x0841, 0x0858, 1},
+ {0x08a0, 0x08b2, 1},
+ {0x0904, 0x0939, 1},
+ {0x093d, 0x0950, 19},
+ {0x0958, 0x0961, 1},
+ {0x0971, 0x0980, 1},
+ {0x0985, 0x098c, 1},
+ {0x098f, 0x0990, 1},
+ {0x0993, 0x09a8, 1},
+ {0x09aa, 0x09b0, 1},
+ {0x09b2, 0x09b6, 4},
+ {0x09b7, 0x09b9, 1},
+ {0x09bd, 0x09ce, 17},
+ {0x09dc, 0x09dd, 1},
+ {0x09df, 0x09e1, 1},
+ {0x09f0, 0x09f1, 1},
+ {0x0a05, 0x0a0a, 1},
+ {0x0a0f, 0x0a10, 1},
+ {0x0a13, 0x0a28, 1},
+ {0x0a2a, 0x0a30, 1},
+ {0x0a32, 0x0a33, 1},
+ {0x0a35, 0x0a36, 1},
+ {0x0a38, 0x0a39, 1},
+ {0x0a59, 0x0a5c, 1},
+ {0x0a5e, 0x0a72, 20},
+ {0x0a73, 0x0a74, 1},
+ {0x0a85, 0x0a8d, 1},
+ {0x0a8f, 0x0a91, 1},
+ {0x0a93, 0x0aa8, 1},
+ {0x0aaa, 0x0ab0, 1},
+ {0x0ab2, 0x0ab3, 1},
+ {0x0ab5, 0x0ab9, 1},
+ {0x0abd, 0x0ad0, 19},
+ {0x0ae0, 0x0ae1, 1},
+ {0x0b05, 0x0b0c, 1},
+ {0x0b0f, 0x0b10, 1},
+ {0x0b13, 0x0b28, 1},
+ {0x0b2a, 0x0b30, 1},
+ {0x0b32, 0x0b33, 1},
+ {0x0b35, 0x0b39, 1},
+ {0x0b3d, 0x0b5c, 31},
+ {0x0b5d, 0x0b5f, 2},
+ {0x0b60, 0x0b61, 1},
+ {0x0b71, 0x0b83, 18},
+ {0x0b85, 0x0b8a, 1},
+ {0x0b8e, 0x0b90, 1},
+ {0x0b92, 0x0b95, 1},
+ {0x0b99, 0x0b9a, 1},
+ {0x0b9c, 0x0b9e, 2},
+ {0x0b9f, 0x0ba3, 4},
+ {0x0ba4, 0x0ba8, 4},
+ {0x0ba9, 0x0baa, 1},
+ {0x0bae, 0x0bb9, 1},
+ {0x0bd0, 0x0c05, 53},
+ {0x0c06, 0x0c0c, 1},
+ {0x0c0e, 0x0c10, 1},
+ {0x0c12, 0x0c28, 1},
+ {0x0c2a, 0x0c39, 1},
+ {0x0c3d, 0x0c58, 27},
+ {0x0c59, 0x0c60, 7},
+ {0x0c61, 0x0c85, 36},
+ {0x0c86, 0x0c8c, 1},
+ {0x0c8e, 0x0c90, 1},
+ {0x0c92, 0x0ca8, 1},
+ {0x0caa, 0x0cb3, 1},
+ {0x0cb5, 0x0cb9, 1},
+ {0x0cbd, 0x0cde, 33},
+ {0x0ce0, 0x0ce1, 1},
+ {0x0cf1, 0x0cf2, 1},
+ {0x0d05, 0x0d0c, 1},
+ {0x0d0e, 0x0d10, 1},
+ {0x0d12, 0x0d3a, 1},
+ {0x0d3d, 0x0d4e, 17},
+ {0x0d60, 0x0d61, 1},
+ {0x0d7a, 0x0d7f, 1},
+ {0x0d85, 0x0d96, 1},
+ {0x0d9a, 0x0db1, 1},
+ {0x0db3, 0x0dbb, 1},
+ {0x0dbd, 0x0dc0, 3},
+ {0x0dc1, 0x0dc6, 1},
+ {0x0e01, 0x0e30, 1},
+ {0x0e32, 0x0e33, 1},
+ {0x0e40, 0x0e46, 1},
+ {0x0e81, 0x0e82, 1},
+ {0x0e84, 0x0e87, 3},
+ {0x0e88, 0x0e8a, 2},
+ {0x0e8d, 0x0e94, 7},
+ {0x0e95, 0x0e97, 1},
+ {0x0e99, 0x0e9f, 1},
+ {0x0ea1, 0x0ea3, 1},
+ {0x0ea5, 0x0ea7, 2},
+ {0x0eaa, 0x0eab, 1},
+ {0x0ead, 0x0eb0, 1},
+ {0x0eb2, 0x0eb3, 1},
+ {0x0ebd, 0x0ec0, 3},
+ {0x0ec1, 0x0ec4, 1},
+ {0x0ec6, 0x0edc, 22},
+ {0x0edd, 0x0edf, 1},
+ {0x0f00, 0x0f40, 64},
+ {0x0f41, 0x0f47, 1},
+ {0x0f49, 0x0f6c, 1},
+ {0x0f88, 0x0f8c, 1},
+ {0x1000, 0x102a, 1},
+ {0x103f, 0x1050, 17},
+ {0x1051, 0x1055, 1},
+ {0x105a, 0x105d, 1},
+ {0x1061, 0x1065, 4},
+ {0x1066, 0x106e, 8},
+ {0x106f, 0x1070, 1},
+ {0x1075, 0x1081, 1},
+ {0x108e, 0x10a0, 18},
+ {0x10a1, 0x10c5, 1},
+ {0x10c7, 0x10cd, 6},
+ {0x10d0, 0x10fa, 1},
+ {0x10fc, 0x1248, 1},
+ {0x124a, 0x124d, 1},
+ {0x1250, 0x1256, 1},
+ {0x1258, 0x125a, 2},
+ {0x125b, 0x125d, 1},
+ {0x1260, 0x1288, 1},
+ {0x128a, 0x128d, 1},
+ {0x1290, 0x12b0, 1},
+ {0x12b2, 0x12b5, 1},
+ {0x12b8, 0x12be, 1},
+ {0x12c0, 0x12c2, 2},
+ {0x12c3, 0x12c5, 1},
+ {0x12c8, 0x12d6, 1},
+ {0x12d8, 0x1310, 1},
+ {0x1312, 0x1315, 1},
+ {0x1318, 0x135a, 1},
+ {0x1380, 0x138f, 1},
+ {0x13a0, 0x13f4, 1},
+ {0x1401, 0x166c, 1},
+ {0x166f, 0x167f, 1},
+ {0x1681, 0x169a, 1},
+ {0x16a0, 0x16ea, 1},
+ {0x16f1, 0x16f8, 1},
+ {0x1700, 0x170c, 1},
+ {0x170e, 0x1711, 1},
+ {0x1720, 0x1731, 1},
+ {0x1740, 0x1751, 1},
+ {0x1760, 0x176c, 1},
+ {0x176e, 0x1770, 1},
+ {0x1780, 0x17b3, 1},
+ {0x17d7, 0x17dc, 5},
+ {0x1820, 0x1877, 1},
+ {0x1880, 0x18a8, 1},
+ {0x18aa, 0x18b0, 6},
+ {0x18b1, 0x18f5, 1},
+ {0x1900, 0x191e, 1},
+ {0x1950, 0x196d, 1},
+ {0x1970, 0x1974, 1},
+ {0x1980, 0x19ab, 1},
+ {0x19c1, 0x19c7, 1},
+ {0x1a00, 0x1a16, 1},
+ {0x1a20, 0x1a54, 1},
+ {0x1aa7, 0x1b05, 94},
+ {0x1b06, 0x1b33, 1},
+ {0x1b45, 0x1b4b, 1},
+ {0x1b83, 0x1ba0, 1},
+ {0x1bae, 0x1baf, 1},
+ {0x1bba, 0x1be5, 1},
+ {0x1c00, 0x1c23, 1},
+ {0x1c4d, 0x1c4f, 1},
+ {0x1c5a, 0x1c7d, 1},
+ {0x1ce9, 0x1cec, 1},
+ {0x1cee, 0x1cf1, 1},
+ {0x1cf5, 0x1cf6, 1},
+ {0x1d00, 0x1dbf, 1},
+ {0x1e00, 0x1f15, 1},
+ {0x1f18, 0x1f1d, 1},
+ {0x1f20, 0x1f45, 1},
+ {0x1f48, 0x1f4d, 1},
+ {0x1f50, 0x1f57, 1},
+ {0x1f59, 0x1f5f, 2},
+ {0x1f60, 0x1f7d, 1},
+ {0x1f80, 0x1fb4, 1},
+ {0x1fb6, 0x1fbc, 1},
+ {0x1fbe, 0x1fc2, 4},
+ {0x1fc3, 0x1fc4, 1},
+ {0x1fc6, 0x1fcc, 1},
+ {0x1fd0, 0x1fd3, 1},
+ {0x1fd6, 0x1fdb, 1},
+ {0x1fe0, 0x1fec, 1},
+ {0x1ff2, 0x1ff4, 1},
+ {0x1ff6, 0x1ffc, 1},
+ {0x2071, 0x207f, 14},
+ {0x2090, 0x209c, 1},
+ {0x2102, 0x2107, 5},
+ {0x210a, 0x2113, 1},
+ {0x2115, 0x2119, 4},
+ {0x211a, 0x211d, 1},
+ {0x2124, 0x212a, 2},
+ {0x212b, 0x212d, 1},
+ {0x212f, 0x2139, 1},
+ {0x213c, 0x213f, 1},
+ {0x2145, 0x2149, 1},
+ {0x214e, 0x2183, 53},
+ {0x2184, 0x2c00, 2684},
+ {0x2c01, 0x2c2e, 1},
+ {0x2c30, 0x2c5e, 1},
+ {0x2c60, 0x2ce4, 1},
+ {0x2ceb, 0x2cee, 1},
+ {0x2cf2, 0x2cf3, 1},
+ {0x2d00, 0x2d25, 1},
+ {0x2d27, 0x2d2d, 6},
+ {0x2d30, 0x2d67, 1},
+ {0x2d6f, 0x2d80, 17},
+ {0x2d81, 0x2d96, 1},
+ {0x2da0, 0x2da6, 1},
+ {0x2da8, 0x2dae, 1},
+ {0x2db0, 0x2db6, 1},
+ {0x2db8, 0x2dbe, 1},
+ {0x2dc0, 0x2dc6, 1},
+ {0x2dc8, 0x2dce, 1},
+ {0x2dd0, 0x2dd6, 1},
+ {0x2dd8, 0x2dde, 1},
+ {0x2e2f, 0x3005, 470},
+ {0x3006, 0x3031, 43},
+ {0x3032, 0x3035, 1},
+ {0x303b, 0x303c, 1},
+ {0x3041, 0x3096, 1},
+ {0x309d, 0x309f, 1},
+ {0x30a1, 0x30fa, 1},
+ {0x30fc, 0x30ff, 1},
+ {0x3105, 0x312d, 1},
+ {0x3131, 0x318e, 1},
+ {0x31a0, 0x31ba, 1},
+ {0x31f0, 0x31ff, 1},
+ {0x3400, 0x4db5, 1},
+ {0x4e00, 0x9fcc, 1},
+ {0xa000, 0xa48c, 1},
+ {0xa4d0, 0xa4fd, 1},
+ {0xa500, 0xa60c, 1},
+ {0xa610, 0xa61f, 1},
+ {0xa62a, 0xa62b, 1},
+ {0xa640, 0xa66e, 1},
+ {0xa67f, 0xa69d, 1},
+ {0xa6a0, 0xa6e5, 1},
+ {0xa717, 0xa71f, 1},
+ {0xa722, 0xa788, 1},
+ {0xa78b, 0xa78e, 1},
+ {0xa790, 0xa7ad, 1},
+ {0xa7b0, 0xa7b1, 1},
+ {0xa7f7, 0xa801, 1},
+ {0xa803, 0xa805, 1},
+ {0xa807, 0xa80a, 1},
+ {0xa80c, 0xa822, 1},
+ {0xa840, 0xa873, 1},
+ {0xa882, 0xa8b3, 1},
+ {0xa8f2, 0xa8f7, 1},
+ {0xa8fb, 0xa90a, 15},
+ {0xa90b, 0xa925, 1},
+ {0xa930, 0xa946, 1},
+ {0xa960, 0xa97c, 1},
+ {0xa984, 0xa9b2, 1},
+ {0xa9cf, 0xa9e0, 17},
+ {0xa9e1, 0xa9e4, 1},
+ {0xa9e6, 0xa9ef, 1},
+ {0xa9fa, 0xa9fe, 1},
+ {0xaa00, 0xaa28, 1},
+ {0xaa40, 0xaa42, 1},
+ {0xaa44, 0xaa4b, 1},
+ {0xaa60, 0xaa76, 1},
+ {0xaa7a, 0xaa7e, 4},
+ {0xaa7f, 0xaaaf, 1},
+ {0xaab1, 0xaab5, 4},
+ {0xaab6, 0xaab9, 3},
+ {0xaaba, 0xaabd, 1},
+ {0xaac0, 0xaac2, 2},
+ {0xaadb, 0xaadd, 1},
+ {0xaae0, 0xaaea, 1},
+ {0xaaf2, 0xaaf4, 1},
+ {0xab01, 0xab06, 1},
+ {0xab09, 0xab0e, 1},
+ {0xab11, 0xab16, 1},
+ {0xab20, 0xab26, 1},
+ {0xab28, 0xab2e, 1},
+ {0xab30, 0xab5a, 1},
+ {0xab5c, 0xab5f, 1},
+ {0xab64, 0xab65, 1},
+ {0xabc0, 0xabe2, 1},
+ {0xac00, 0xd7a3, 1},
+ {0xd7b0, 0xd7c6, 1},
+ {0xd7cb, 0xd7fb, 1},
+ {0xf900, 0xfa6d, 1},
+ {0xfa70, 0xfad9, 1},
+ {0xfb00, 0xfb06, 1},
+ {0xfb13, 0xfb17, 1},
+ {0xfb1d, 0xfb1f, 2},
+ {0xfb20, 0xfb28, 1},
+ {0xfb2a, 0xfb36, 1},
+ {0xfb38, 0xfb3c, 1},
+ {0xfb3e, 0xfb40, 2},
+ {0xfb41, 0xfb43, 2},
+ {0xfb44, 0xfb46, 2},
+ {0xfb47, 0xfbb1, 1},
+ {0xfbd3, 0xfd3d, 1},
+ {0xfd50, 0xfd8f, 1},
+ {0xfd92, 0xfdc7, 1},
+ {0xfdf0, 0xfdfb, 1},
+ {0xfe70, 0xfe74, 1},
+ {0xfe76, 0xfefc, 1},
+ {0xff21, 0xff3a, 1},
+ {0xff41, 0xff5a, 1},
+ {0xff66, 0xffbe, 1},
+ {0xffc2, 0xffc7, 1},
+ {0xffca, 0xffcf, 1},
+ {0xffd2, 0xffd7, 1},
+ {0xffda, 0xffdc, 1},
+ },
+ R32: []Range32{
+ {0x10000, 0x1000b, 1},
+ {0x1000d, 0x10026, 1},
+ {0x10028, 0x1003a, 1},
+ {0x1003c, 0x1003d, 1},
+ {0x1003f, 0x1004d, 1},
+ {0x10050, 0x1005d, 1},
+ {0x10080, 0x100fa, 1},
+ {0x10280, 0x1029c, 1},
+ {0x102a0, 0x102d0, 1},
+ {0x10300, 0x1031f, 1},
+ {0x10330, 0x10340, 1},
+ {0x10342, 0x10349, 1},
+ {0x10350, 0x10375, 1},
+ {0x10380, 0x1039d, 1},
+ {0x103a0, 0x103c3, 1},
+ {0x103c8, 0x103cf, 1},
+ {0x10400, 0x1049d, 1},
+ {0x10500, 0x10527, 1},
+ {0x10530, 0x10563, 1},
+ {0x10600, 0x10736, 1},
+ {0x10740, 0x10755, 1},
+ {0x10760, 0x10767, 1},
+ {0x10800, 0x10805, 1},
+ {0x10808, 0x1080a, 2},
+ {0x1080b, 0x10835, 1},
+ {0x10837, 0x10838, 1},
+ {0x1083c, 0x1083f, 3},
+ {0x10840, 0x10855, 1},
+ {0x10860, 0x10876, 1},
+ {0x10880, 0x1089e, 1},
+ {0x10900, 0x10915, 1},
+ {0x10920, 0x10939, 1},
+ {0x10980, 0x109b7, 1},
+ {0x109be, 0x109bf, 1},
+ {0x10a00, 0x10a10, 16},
+ {0x10a11, 0x10a13, 1},
+ {0x10a15, 0x10a17, 1},
+ {0x10a19, 0x10a33, 1},
+ {0x10a60, 0x10a7c, 1},
+ {0x10a80, 0x10a9c, 1},
+ {0x10ac0, 0x10ac7, 1},
+ {0x10ac9, 0x10ae4, 1},
+ {0x10b00, 0x10b35, 1},
+ {0x10b40, 0x10b55, 1},
+ {0x10b60, 0x10b72, 1},
+ {0x10b80, 0x10b91, 1},
+ {0x10c00, 0x10c48, 1},
+ {0x11003, 0x11037, 1},
+ {0x11083, 0x110af, 1},
+ {0x110d0, 0x110e8, 1},
+ {0x11103, 0x11126, 1},
+ {0x11150, 0x11172, 1},
+ {0x11176, 0x11183, 13},
+ {0x11184, 0x111b2, 1},
+ {0x111c1, 0x111c4, 1},
+ {0x111da, 0x11200, 38},
+ {0x11201, 0x11211, 1},
+ {0x11213, 0x1122b, 1},
+ {0x112b0, 0x112de, 1},
+ {0x11305, 0x1130c, 1},
+ {0x1130f, 0x11310, 1},
+ {0x11313, 0x11328, 1},
+ {0x1132a, 0x11330, 1},
+ {0x11332, 0x11333, 1},
+ {0x11335, 0x11339, 1},
+ {0x1133d, 0x1135d, 32},
+ {0x1135e, 0x11361, 1},
+ {0x11480, 0x114af, 1},
+ {0x114c4, 0x114c5, 1},
+ {0x114c7, 0x11580, 185},
+ {0x11581, 0x115ae, 1},
+ {0x11600, 0x1162f, 1},
+ {0x11644, 0x11680, 60},
+ {0x11681, 0x116aa, 1},
+ {0x118a0, 0x118df, 1},
+ {0x118ff, 0x11ac0, 449},
+ {0x11ac1, 0x11af8, 1},
+ {0x12000, 0x12398, 1},
+ {0x13000, 0x1342e, 1},
+ {0x16800, 0x16a38, 1},
+ {0x16a40, 0x16a5e, 1},
+ {0x16ad0, 0x16aed, 1},
+ {0x16b00, 0x16b2f, 1},
+ {0x16b40, 0x16b43, 1},
+ {0x16b63, 0x16b77, 1},
+ {0x16b7d, 0x16b8f, 1},
+ {0x16f00, 0x16f44, 1},
+ {0x16f50, 0x16f93, 67},
+ {0x16f94, 0x16f9f, 1},
+ {0x1b000, 0x1b001, 1},
+ {0x1bc00, 0x1bc6a, 1},
+ {0x1bc70, 0x1bc7c, 1},
+ {0x1bc80, 0x1bc88, 1},
+ {0x1bc90, 0x1bc99, 1},
+ {0x1d400, 0x1d454, 1},
+ {0x1d456, 0x1d49c, 1},
+ {0x1d49e, 0x1d49f, 1},
+ {0x1d4a2, 0x1d4a5, 3},
+ {0x1d4a6, 0x1d4a9, 3},
+ {0x1d4aa, 0x1d4ac, 1},
+ {0x1d4ae, 0x1d4b9, 1},
+ {0x1d4bb, 0x1d4bd, 2},
+ {0x1d4be, 0x1d4c3, 1},
+ {0x1d4c5, 0x1d505, 1},
+ {0x1d507, 0x1d50a, 1},
+ {0x1d50d, 0x1d514, 1},
+ {0x1d516, 0x1d51c, 1},
+ {0x1d51e, 0x1d539, 1},
+ {0x1d53b, 0x1d53e, 1},
+ {0x1d540, 0x1d544, 1},
+ {0x1d546, 0x1d54a, 4},
+ {0x1d54b, 0x1d550, 1},
+ {0x1d552, 0x1d6a5, 1},
+ {0x1d6a8, 0x1d6c0, 1},
+ {0x1d6c2, 0x1d6da, 1},
+ {0x1d6dc, 0x1d6fa, 1},
+ {0x1d6fc, 0x1d714, 1},
+ {0x1d716, 0x1d734, 1},
+ {0x1d736, 0x1d74e, 1},
+ {0x1d750, 0x1d76e, 1},
+ {0x1d770, 0x1d788, 1},
+ {0x1d78a, 0x1d7a8, 1},
+ {0x1d7aa, 0x1d7c2, 1},
+ {0x1d7c4, 0x1d7cb, 1},
+ {0x1e800, 0x1e8c4, 1},
+ {0x1ee00, 0x1ee03, 1},
+ {0x1ee05, 0x1ee1f, 1},
+ {0x1ee21, 0x1ee22, 1},
+ {0x1ee24, 0x1ee27, 3},
+ {0x1ee29, 0x1ee32, 1},
+ {0x1ee34, 0x1ee37, 1},
+ {0x1ee39, 0x1ee3b, 2},
+ {0x1ee42, 0x1ee47, 5},
+ {0x1ee49, 0x1ee4d, 2},
+ {0x1ee4e, 0x1ee4f, 1},
+ {0x1ee51, 0x1ee52, 1},
+ {0x1ee54, 0x1ee57, 3},
+ {0x1ee59, 0x1ee61, 2},
+ {0x1ee62, 0x1ee64, 2},
+ {0x1ee67, 0x1ee6a, 1},
+ {0x1ee6c, 0x1ee72, 1},
+ {0x1ee74, 0x1ee77, 1},
+ {0x1ee79, 0x1ee7c, 1},
+ {0x1ee7e, 0x1ee80, 2},
+ {0x1ee81, 0x1ee89, 1},
+ {0x1ee8b, 0x1ee9b, 1},
+ {0x1eea1, 0x1eea3, 1},
+ {0x1eea5, 0x1eea9, 1},
+ {0x1eeab, 0x1eebb, 1},
+ {0x20000, 0x2a6d6, 1},
+ {0x2a700, 0x2b734, 1},
+ {0x2b740, 0x2b81d, 1},
+ {0x2f800, 0x2fa1d, 1},
+ },
+ LatinOffset: 6,
+}
+
+var _Ll = &RangeTable{
+ R16: []Range16{
+ {0x0061, 0x007a, 1},
+ {0x00b5, 0x00df, 42},
+ {0x00e0, 0x00f6, 1},
+ {0x00f8, 0x00ff, 1},
+ {0x0101, 0x0137, 2},
+ {0x0138, 0x0148, 2},
+ {0x0149, 0x0177, 2},
+ {0x017a, 0x017e, 2},
+ {0x017f, 0x0180, 1},
+ {0x0183, 0x0185, 2},
+ {0x0188, 0x018c, 4},
+ {0x018d, 0x0192, 5},
+ {0x0195, 0x0199, 4},
+ {0x019a, 0x019b, 1},
+ {0x019e, 0x01a1, 3},
+ {0x01a3, 0x01a5, 2},
+ {0x01a8, 0x01aa, 2},
+ {0x01ab, 0x01ad, 2},
+ {0x01b0, 0x01b4, 4},
+ {0x01b6, 0x01b9, 3},
+ {0x01ba, 0x01bd, 3},
+ {0x01be, 0x01bf, 1},
+ {0x01c6, 0x01cc, 3},
+ {0x01ce, 0x01dc, 2},
+ {0x01dd, 0x01ef, 2},
+ {0x01f0, 0x01f3, 3},
+ {0x01f5, 0x01f9, 4},
+ {0x01fb, 0x0233, 2},
+ {0x0234, 0x0239, 1},
+ {0x023c, 0x023f, 3},
+ {0x0240, 0x0242, 2},
+ {0x0247, 0x024f, 2},
+ {0x0250, 0x0293, 1},
+ {0x0295, 0x02af, 1},
+ {0x0371, 0x0373, 2},
+ {0x0377, 0x037b, 4},
+ {0x037c, 0x037d, 1},
+ {0x0390, 0x03ac, 28},
+ {0x03ad, 0x03ce, 1},
+ {0x03d0, 0x03d1, 1},
+ {0x03d5, 0x03d7, 1},
+ {0x03d9, 0x03ef, 2},
+ {0x03f0, 0x03f3, 1},
+ {0x03f5, 0x03fb, 3},
+ {0x03fc, 0x0430, 52},
+ {0x0431, 0x045f, 1},
+ {0x0461, 0x0481, 2},
+ {0x048b, 0x04bf, 2},
+ {0x04c2, 0x04ce, 2},
+ {0x04cf, 0x052f, 2},
+ {0x0561, 0x0587, 1},
+ {0x1d00, 0x1d2b, 1},
+ {0x1d6b, 0x1d77, 1},
+ {0x1d79, 0x1d9a, 1},
+ {0x1e01, 0x1e95, 2},
+ {0x1e96, 0x1e9d, 1},
+ {0x1e9f, 0x1eff, 2},
+ {0x1f00, 0x1f07, 1},
+ {0x1f10, 0x1f15, 1},
+ {0x1f20, 0x1f27, 1},
+ {0x1f30, 0x1f37, 1},
+ {0x1f40, 0x1f45, 1},
+ {0x1f50, 0x1f57, 1},
+ {0x1f60, 0x1f67, 1},
+ {0x1f70, 0x1f7d, 1},
+ {0x1f80, 0x1f87, 1},
+ {0x1f90, 0x1f97, 1},
+ {0x1fa0, 0x1fa7, 1},
+ {0x1fb0, 0x1fb4, 1},
+ {0x1fb6, 0x1fb7, 1},
+ {0x1fbe, 0x1fc2, 4},
+ {0x1fc3, 0x1fc4, 1},
+ {0x1fc6, 0x1fc7, 1},
+ {0x1fd0, 0x1fd3, 1},
+ {0x1fd6, 0x1fd7, 1},
+ {0x1fe0, 0x1fe7, 1},
+ {0x1ff2, 0x1ff4, 1},
+ {0x1ff6, 0x1ff7, 1},
+ {0x210a, 0x210e, 4},
+ {0x210f, 0x2113, 4},
+ {0x212f, 0x2139, 5},
+ {0x213c, 0x213d, 1},
+ {0x2146, 0x2149, 1},
+ {0x214e, 0x2184, 54},
+ {0x2c30, 0x2c5e, 1},
+ {0x2c61, 0x2c65, 4},
+ {0x2c66, 0x2c6c, 2},
+ {0x2c71, 0x2c73, 2},
+ {0x2c74, 0x2c76, 2},
+ {0x2c77, 0x2c7b, 1},
+ {0x2c81, 0x2ce3, 2},
+ {0x2ce4, 0x2cec, 8},
+ {0x2cee, 0x2cf3, 5},
+ {0x2d00, 0x2d25, 1},
+ {0x2d27, 0x2d2d, 6},
+ {0xa641, 0xa66d, 2},
+ {0xa681, 0xa69b, 2},
+ {0xa723, 0xa72f, 2},
+ {0xa730, 0xa731, 1},
+ {0xa733, 0xa771, 2},
+ {0xa772, 0xa778, 1},
+ {0xa77a, 0xa77c, 2},
+ {0xa77f, 0xa787, 2},
+ {0xa78c, 0xa78e, 2},
+ {0xa791, 0xa793, 2},
+ {0xa794, 0xa795, 1},
+ {0xa797, 0xa7a9, 2},
+ {0xa7fa, 0xab30, 822},
+ {0xab31, 0xab5a, 1},
+ {0xab64, 0xab65, 1},
+ {0xfb00, 0xfb06, 1},
+ {0xfb13, 0xfb17, 1},
+ {0xff41, 0xff5a, 1},
+ },
+ R32: []Range32{
+ {0x10428, 0x1044f, 1},
+ {0x118c0, 0x118df, 1},
+ {0x1d41a, 0x1d433, 1},
+ {0x1d44e, 0x1d454, 1},
+ {0x1d456, 0x1d467, 1},
+ {0x1d482, 0x1d49b, 1},
+ {0x1d4b6, 0x1d4b9, 1},
+ {0x1d4bb, 0x1d4bd, 2},
+ {0x1d4be, 0x1d4c3, 1},
+ {0x1d4c5, 0x1d4cf, 1},
+ {0x1d4ea, 0x1d503, 1},
+ {0x1d51e, 0x1d537, 1},
+ {0x1d552, 0x1d56b, 1},
+ {0x1d586, 0x1d59f, 1},
+ {0x1d5ba, 0x1d5d3, 1},
+ {0x1d5ee, 0x1d607, 1},
+ {0x1d622, 0x1d63b, 1},
+ {0x1d656, 0x1d66f, 1},
+ {0x1d68a, 0x1d6a5, 1},
+ {0x1d6c2, 0x1d6da, 1},
+ {0x1d6dc, 0x1d6e1, 1},
+ {0x1d6fc, 0x1d714, 1},
+ {0x1d716, 0x1d71b, 1},
+ {0x1d736, 0x1d74e, 1},
+ {0x1d750, 0x1d755, 1},
+ {0x1d770, 0x1d788, 1},
+ {0x1d78a, 0x1d78f, 1},
+ {0x1d7aa, 0x1d7c2, 1},
+ {0x1d7c4, 0x1d7c9, 1},
+ {0x1d7cb, 0x1d7cb, 1},
+ },
+ LatinOffset: 4,
+}
+
+var _Lm = &RangeTable{
+ R16: []Range16{
+ {0x02b0, 0x02c1, 1},
+ {0x02c6, 0x02d1, 1},
+ {0x02e0, 0x02e4, 1},
+ {0x02ec, 0x02ee, 2},
+ {0x0374, 0x037a, 6},
+ {0x0559, 0x0640, 231},
+ {0x06e5, 0x06e6, 1},
+ {0x07f4, 0x07f5, 1},
+ {0x07fa, 0x081a, 32},
+ {0x0824, 0x0828, 4},
+ {0x0971, 0x0e46, 1237},
+ {0x0ec6, 0x10fc, 566},
+ {0x17d7, 0x1843, 108},
+ {0x1aa7, 0x1c78, 465},
+ {0x1c79, 0x1c7d, 1},
+ {0x1d2c, 0x1d6a, 1},
+ {0x1d78, 0x1d9b, 35},
+ {0x1d9c, 0x1dbf, 1},
+ {0x2071, 0x207f, 14},
+ {0x2090, 0x209c, 1},
+ {0x2c7c, 0x2c7d, 1},
+ {0x2d6f, 0x2e2f, 192},
+ {0x3005, 0x3031, 44},
+ {0x3032, 0x3035, 1},
+ {0x303b, 0x309d, 98},
+ {0x309e, 0x30fc, 94},
+ {0x30fd, 0x30fe, 1},
+ {0xa015, 0xa4f8, 1251},
+ {0xa4f9, 0xa4fd, 1},
+ {0xa60c, 0xa67f, 115},
+ {0xa69c, 0xa69d, 1},
+ {0xa717, 0xa71f, 1},
+ {0xa770, 0xa788, 24},
+ {0xa7f8, 0xa7f9, 1},
+ {0xa9cf, 0xa9e6, 23},
+ {0xaa70, 0xaadd, 109},
+ {0xaaf3, 0xaaf4, 1},
+ {0xab5c, 0xab5f, 1},
+ {0xff70, 0xff9e, 46},
+ {0xff9f, 0xff9f, 1},
+ },
+ R32: []Range32{
+ {0x16b40, 0x16b40, 1},
+ {0x16b41, 0x16b43, 1},
+ {0x16f93, 0x16f9f, 1},
+ },
+}
+
+var _Lo = &RangeTable{
+ R16: []Range16{
+ {0x00aa, 0x00ba, 16},
+ {0x01bb, 0x01c0, 5},
+ {0x01c1, 0x01c3, 1},
+ {0x0294, 0x05d0, 828},
+ {0x05d1, 0x05ea, 1},
+ {0x05f0, 0x05f2, 1},
+ {0x0620, 0x063f, 1},
+ {0x0641, 0x064a, 1},
+ {0x066e, 0x066f, 1},
+ {0x0671, 0x06d3, 1},
+ {0x06d5, 0x06ee, 25},
+ {0x06ef, 0x06fa, 11},
+ {0x06fb, 0x06fc, 1},
+ {0x06ff, 0x0710, 17},
+ {0x0712, 0x072f, 1},
+ {0x074d, 0x07a5, 1},
+ {0x07b1, 0x07ca, 25},
+ {0x07cb, 0x07ea, 1},
+ {0x0800, 0x0815, 1},
+ {0x0840, 0x0858, 1},
+ {0x08a0, 0x08b2, 1},
+ {0x0904, 0x0939, 1},
+ {0x093d, 0x0950, 19},
+ {0x0958, 0x0961, 1},
+ {0x0972, 0x0980, 1},
+ {0x0985, 0x098c, 1},
+ {0x098f, 0x0990, 1},
+ {0x0993, 0x09a8, 1},
+ {0x09aa, 0x09b0, 1},
+ {0x09b2, 0x09b6, 4},
+ {0x09b7, 0x09b9, 1},
+ {0x09bd, 0x09ce, 17},
+ {0x09dc, 0x09dd, 1},
+ {0x09df, 0x09e1, 1},
+ {0x09f0, 0x09f1, 1},
+ {0x0a05, 0x0a0a, 1},
+ {0x0a0f, 0x0a10, 1},
+ {0x0a13, 0x0a28, 1},
+ {0x0a2a, 0x0a30, 1},
+ {0x0a32, 0x0a33, 1},
+ {0x0a35, 0x0a36, 1},
+ {0x0a38, 0x0a39, 1},
+ {0x0a59, 0x0a5c, 1},
+ {0x0a5e, 0x0a72, 20},
+ {0x0a73, 0x0a74, 1},
+ {0x0a85, 0x0a8d, 1},
+ {0x0a8f, 0x0a91, 1},
+ {0x0a93, 0x0aa8, 1},
+ {0x0aaa, 0x0ab0, 1},
+ {0x0ab2, 0x0ab3, 1},
+ {0x0ab5, 0x0ab9, 1},
+ {0x0abd, 0x0ad0, 19},
+ {0x0ae0, 0x0ae1, 1},
+ {0x0b05, 0x0b0c, 1},
+ {0x0b0f, 0x0b10, 1},
+ {0x0b13, 0x0b28, 1},
+ {0x0b2a, 0x0b30, 1},
+ {0x0b32, 0x0b33, 1},
+ {0x0b35, 0x0b39, 1},
+ {0x0b3d, 0x0b5c, 31},
+ {0x0b5d, 0x0b5f, 2},
+ {0x0b60, 0x0b61, 1},
+ {0x0b71, 0x0b83, 18},
+ {0x0b85, 0x0b8a, 1},
+ {0x0b8e, 0x0b90, 1},
+ {0x0b92, 0x0b95, 1},
+ {0x0b99, 0x0b9a, 1},
+ {0x0b9c, 0x0b9e, 2},
+ {0x0b9f, 0x0ba3, 4},
+ {0x0ba4, 0x0ba8, 4},
+ {0x0ba9, 0x0baa, 1},
+ {0x0bae, 0x0bb9, 1},
+ {0x0bd0, 0x0c05, 53},
+ {0x0c06, 0x0c0c, 1},
+ {0x0c0e, 0x0c10, 1},
+ {0x0c12, 0x0c28, 1},
+ {0x0c2a, 0x0c39, 1},
+ {0x0c3d, 0x0c58, 27},
+ {0x0c59, 0x0c60, 7},
+ {0x0c61, 0x0c85, 36},
+ {0x0c86, 0x0c8c, 1},
+ {0x0c8e, 0x0c90, 1},
+ {0x0c92, 0x0ca8, 1},
+ {0x0caa, 0x0cb3, 1},
+ {0x0cb5, 0x0cb9, 1},
+ {0x0cbd, 0x0cde, 33},
+ {0x0ce0, 0x0ce1, 1},
+ {0x0cf1, 0x0cf2, 1},
+ {0x0d05, 0x0d0c, 1},
+ {0x0d0e, 0x0d10, 1},
+ {0x0d12, 0x0d3a, 1},
+ {0x0d3d, 0x0d4e, 17},
+ {0x0d60, 0x0d61, 1},
+ {0x0d7a, 0x0d7f, 1},
+ {0x0d85, 0x0d96, 1},
+ {0x0d9a, 0x0db1, 1},
+ {0x0db3, 0x0dbb, 1},
+ {0x0dbd, 0x0dc0, 3},
+ {0x0dc1, 0x0dc6, 1},
+ {0x0e01, 0x0e30, 1},
+ {0x0e32, 0x0e33, 1},
+ {0x0e40, 0x0e45, 1},
+ {0x0e81, 0x0e82, 1},
+ {0x0e84, 0x0e87, 3},
+ {0x0e88, 0x0e8a, 2},
+ {0x0e8d, 0x0e94, 7},
+ {0x0e95, 0x0e97, 1},
+ {0x0e99, 0x0e9f, 1},
+ {0x0ea1, 0x0ea3, 1},
+ {0x0ea5, 0x0ea7, 2},
+ {0x0eaa, 0x0eab, 1},
+ {0x0ead, 0x0eb0, 1},
+ {0x0eb2, 0x0eb3, 1},
+ {0x0ebd, 0x0ec0, 3},
+ {0x0ec1, 0x0ec4, 1},
+ {0x0edc, 0x0edf, 1},
+ {0x0f00, 0x0f40, 64},
+ {0x0f41, 0x0f47, 1},
+ {0x0f49, 0x0f6c, 1},
+ {0x0f88, 0x0f8c, 1},
+ {0x1000, 0x102a, 1},
+ {0x103f, 0x1050, 17},
+ {0x1051, 0x1055, 1},
+ {0x105a, 0x105d, 1},
+ {0x1061, 0x1065, 4},
+ {0x1066, 0x106e, 8},
+ {0x106f, 0x1070, 1},
+ {0x1075, 0x1081, 1},
+ {0x108e, 0x10d0, 66},
+ {0x10d1, 0x10fa, 1},
+ {0x10fd, 0x1248, 1},
+ {0x124a, 0x124d, 1},
+ {0x1250, 0x1256, 1},
+ {0x1258, 0x125a, 2},
+ {0x125b, 0x125d, 1},
+ {0x1260, 0x1288, 1},
+ {0x128a, 0x128d, 1},
+ {0x1290, 0x12b0, 1},
+ {0x12b2, 0x12b5, 1},
+ {0x12b8, 0x12be, 1},
+ {0x12c0, 0x12c2, 2},
+ {0x12c3, 0x12c5, 1},
+ {0x12c8, 0x12d6, 1},
+ {0x12d8, 0x1310, 1},
+ {0x1312, 0x1315, 1},
+ {0x1318, 0x135a, 1},
+ {0x1380, 0x138f, 1},
+ {0x13a0, 0x13f4, 1},
+ {0x1401, 0x166c, 1},
+ {0x166f, 0x167f, 1},
+ {0x1681, 0x169a, 1},
+ {0x16a0, 0x16ea, 1},
+ {0x16f1, 0x16f8, 1},
+ {0x1700, 0x170c, 1},
+ {0x170e, 0x1711, 1},
+ {0x1720, 0x1731, 1},
+ {0x1740, 0x1751, 1},
+ {0x1760, 0x176c, 1},
+ {0x176e, 0x1770, 1},
+ {0x1780, 0x17b3, 1},
+ {0x17dc, 0x1820, 68},
+ {0x1821, 0x1842, 1},
+ {0x1844, 0x1877, 1},
+ {0x1880, 0x18a8, 1},
+ {0x18aa, 0x18b0, 6},
+ {0x18b1, 0x18f5, 1},
+ {0x1900, 0x191e, 1},
+ {0x1950, 0x196d, 1},
+ {0x1970, 0x1974, 1},
+ {0x1980, 0x19ab, 1},
+ {0x19c1, 0x19c7, 1},
+ {0x1a00, 0x1a16, 1},
+ {0x1a20, 0x1a54, 1},
+ {0x1b05, 0x1b33, 1},
+ {0x1b45, 0x1b4b, 1},
+ {0x1b83, 0x1ba0, 1},
+ {0x1bae, 0x1baf, 1},
+ {0x1bba, 0x1be5, 1},
+ {0x1c00, 0x1c23, 1},
+ {0x1c4d, 0x1c4f, 1},
+ {0x1c5a, 0x1c77, 1},
+ {0x1ce9, 0x1cec, 1},
+ {0x1cee, 0x1cf1, 1},
+ {0x1cf5, 0x1cf6, 1},
+ {0x2135, 0x2138, 1},
+ {0x2d30, 0x2d67, 1},
+ {0x2d80, 0x2d96, 1},
+ {0x2da0, 0x2da6, 1},
+ {0x2da8, 0x2dae, 1},
+ {0x2db0, 0x2db6, 1},
+ {0x2db8, 0x2dbe, 1},
+ {0x2dc0, 0x2dc6, 1},
+ {0x2dc8, 0x2dce, 1},
+ {0x2dd0, 0x2dd6, 1},
+ {0x2dd8, 0x2dde, 1},
+ {0x3006, 0x303c, 54},
+ {0x3041, 0x3096, 1},
+ {0x309f, 0x30a1, 2},
+ {0x30a2, 0x30fa, 1},
+ {0x30ff, 0x3105, 6},
+ {0x3106, 0x312d, 1},
+ {0x3131, 0x318e, 1},
+ {0x31a0, 0x31ba, 1},
+ {0x31f0, 0x31ff, 1},
+ {0x3400, 0x4db5, 1},
+ {0x4e00, 0x9fcc, 1},
+ {0xa000, 0xa014, 1},
+ {0xa016, 0xa48c, 1},
+ {0xa4d0, 0xa4f7, 1},
+ {0xa500, 0xa60b, 1},
+ {0xa610, 0xa61f, 1},
+ {0xa62a, 0xa62b, 1},
+ {0xa66e, 0xa6a0, 50},
+ {0xa6a1, 0xa6e5, 1},
+ {0xa7f7, 0xa7fb, 4},
+ {0xa7fc, 0xa801, 1},
+ {0xa803, 0xa805, 1},
+ {0xa807, 0xa80a, 1},
+ {0xa80c, 0xa822, 1},
+ {0xa840, 0xa873, 1},
+ {0xa882, 0xa8b3, 1},
+ {0xa8f2, 0xa8f7, 1},
+ {0xa8fb, 0xa90a, 15},
+ {0xa90b, 0xa925, 1},
+ {0xa930, 0xa946, 1},
+ {0xa960, 0xa97c, 1},
+ {0xa984, 0xa9b2, 1},
+ {0xa9e0, 0xa9e4, 1},
+ {0xa9e7, 0xa9ef, 1},
+ {0xa9fa, 0xa9fe, 1},
+ {0xaa00, 0xaa28, 1},
+ {0xaa40, 0xaa42, 1},
+ {0xaa44, 0xaa4b, 1},
+ {0xaa60, 0xaa6f, 1},
+ {0xaa71, 0xaa76, 1},
+ {0xaa7a, 0xaa7e, 4},
+ {0xaa7f, 0xaaaf, 1},
+ {0xaab1, 0xaab5, 4},
+ {0xaab6, 0xaab9, 3},
+ {0xaaba, 0xaabd, 1},
+ {0xaac0, 0xaac2, 2},
+ {0xaadb, 0xaadc, 1},
+ {0xaae0, 0xaaea, 1},
+ {0xaaf2, 0xab01, 15},
+ {0xab02, 0xab06, 1},
+ {0xab09, 0xab0e, 1},
+ {0xab11, 0xab16, 1},
+ {0xab20, 0xab26, 1},
+ {0xab28, 0xab2e, 1},
+ {0xabc0, 0xabe2, 1},
+ {0xac00, 0xd7a3, 1},
+ {0xd7b0, 0xd7c6, 1},
+ {0xd7cb, 0xd7fb, 1},
+ {0xf900, 0xfa6d, 1},
+ {0xfa70, 0xfad9, 1},
+ {0xfb1d, 0xfb1f, 2},
+ {0xfb20, 0xfb28, 1},
+ {0xfb2a, 0xfb36, 1},
+ {0xfb38, 0xfb3c, 1},
+ {0xfb3e, 0xfb40, 2},
+ {0xfb41, 0xfb43, 2},
+ {0xfb44, 0xfb46, 2},
+ {0xfb47, 0xfbb1, 1},
+ {0xfbd3, 0xfd3d, 1},
+ {0xfd50, 0xfd8f, 1},
+ {0xfd92, 0xfdc7, 1},
+ {0xfdf0, 0xfdfb, 1},
+ {0xfe70, 0xfe74, 1},
+ {0xfe76, 0xfefc, 1},
+ {0xff66, 0xff6f, 1},
+ {0xff71, 0xff9d, 1},
+ {0xffa0, 0xffbe, 1},
+ {0xffc2, 0xffc7, 1},
+ {0xffca, 0xffcf, 1},
+ {0xffd2, 0xffd7, 1},
+ {0xffda, 0xffdc, 1},
+ },
+ R32: []Range32{
+ {0x10000, 0x1000b, 1},
+ {0x1000d, 0x10026, 1},
+ {0x10028, 0x1003a, 1},
+ {0x1003c, 0x1003d, 1},
+ {0x1003f, 0x1004d, 1},
+ {0x10050, 0x1005d, 1},
+ {0x10080, 0x100fa, 1},
+ {0x10280, 0x1029c, 1},
+ {0x102a0, 0x102d0, 1},
+ {0x10300, 0x1031f, 1},
+ {0x10330, 0x10340, 1},
+ {0x10342, 0x10349, 1},
+ {0x10350, 0x10375, 1},
+ {0x10380, 0x1039d, 1},
+ {0x103a0, 0x103c3, 1},
+ {0x103c8, 0x103cf, 1},
+ {0x10450, 0x1049d, 1},
+ {0x10500, 0x10527, 1},
+ {0x10530, 0x10563, 1},
+ {0x10600, 0x10736, 1},
+ {0x10740, 0x10755, 1},
+ {0x10760, 0x10767, 1},
+ {0x10800, 0x10805, 1},
+ {0x10808, 0x1080a, 2},
+ {0x1080b, 0x10835, 1},
+ {0x10837, 0x10838, 1},
+ {0x1083c, 0x1083f, 3},
+ {0x10840, 0x10855, 1},
+ {0x10860, 0x10876, 1},
+ {0x10880, 0x1089e, 1},
+ {0x10900, 0x10915, 1},
+ {0x10920, 0x10939, 1},
+ {0x10980, 0x109b7, 1},
+ {0x109be, 0x109bf, 1},
+ {0x10a00, 0x10a10, 16},
+ {0x10a11, 0x10a13, 1},
+ {0x10a15, 0x10a17, 1},
+ {0x10a19, 0x10a33, 1},
+ {0x10a60, 0x10a7c, 1},
+ {0x10a80, 0x10a9c, 1},
+ {0x10ac0, 0x10ac7, 1},
+ {0x10ac9, 0x10ae4, 1},
+ {0x10b00, 0x10b35, 1},
+ {0x10b40, 0x10b55, 1},
+ {0x10b60, 0x10b72, 1},
+ {0x10b80, 0x10b91, 1},
+ {0x10c00, 0x10c48, 1},
+ {0x11003, 0x11037, 1},
+ {0x11083, 0x110af, 1},
+ {0x110d0, 0x110e8, 1},
+ {0x11103, 0x11126, 1},
+ {0x11150, 0x11172, 1},
+ {0x11176, 0x11183, 13},
+ {0x11184, 0x111b2, 1},
+ {0x111c1, 0x111c4, 1},
+ {0x111da, 0x11200, 38},
+ {0x11201, 0x11211, 1},
+ {0x11213, 0x1122b, 1},
+ {0x112b0, 0x112de, 1},
+ {0x11305, 0x1130c, 1},
+ {0x1130f, 0x11310, 1},
+ {0x11313, 0x11328, 1},
+ {0x1132a, 0x11330, 1},
+ {0x11332, 0x11333, 1},
+ {0x11335, 0x11339, 1},
+ {0x1133d, 0x1135d, 32},
+ {0x1135e, 0x11361, 1},
+ {0x11480, 0x114af, 1},
+ {0x114c4, 0x114c5, 1},
+ {0x114c7, 0x11580, 185},
+ {0x11581, 0x115ae, 1},
+ {0x11600, 0x1162f, 1},
+ {0x11644, 0x11680, 60},
+ {0x11681, 0x116aa, 1},
+ {0x118ff, 0x11ac0, 449},
+ {0x11ac1, 0x11af8, 1},
+ {0x12000, 0x12398, 1},
+ {0x13000, 0x1342e, 1},
+ {0x16800, 0x16a38, 1},
+ {0x16a40, 0x16a5e, 1},
+ {0x16ad0, 0x16aed, 1},
+ {0x16b00, 0x16b2f, 1},
+ {0x16b63, 0x16b77, 1},
+ {0x16b7d, 0x16b8f, 1},
+ {0x16f00, 0x16f44, 1},
+ {0x16f50, 0x1b000, 16560},
+ {0x1b001, 0x1bc00, 3071},
+ {0x1bc01, 0x1bc6a, 1},
+ {0x1bc70, 0x1bc7c, 1},
+ {0x1bc80, 0x1bc88, 1},
+ {0x1bc90, 0x1bc99, 1},
+ {0x1e800, 0x1e8c4, 1},
+ {0x1ee00, 0x1ee03, 1},
+ {0x1ee05, 0x1ee1f, 1},
+ {0x1ee21, 0x1ee22, 1},
+ {0x1ee24, 0x1ee27, 3},
+ {0x1ee29, 0x1ee32, 1},
+ {0x1ee34, 0x1ee37, 1},
+ {0x1ee39, 0x1ee3b, 2},
+ {0x1ee42, 0x1ee47, 5},
+ {0x1ee49, 0x1ee4d, 2},
+ {0x1ee4e, 0x1ee4f, 1},
+ {0x1ee51, 0x1ee52, 1},
+ {0x1ee54, 0x1ee57, 3},
+ {0x1ee59, 0x1ee61, 2},
+ {0x1ee62, 0x1ee64, 2},
+ {0x1ee67, 0x1ee6a, 1},
+ {0x1ee6c, 0x1ee72, 1},
+ {0x1ee74, 0x1ee77, 1},
+ {0x1ee79, 0x1ee7c, 1},
+ {0x1ee7e, 0x1ee80, 2},
+ {0x1ee81, 0x1ee89, 1},
+ {0x1ee8b, 0x1ee9b, 1},
+ {0x1eea1, 0x1eea3, 1},
+ {0x1eea5, 0x1eea9, 1},
+ {0x1eeab, 0x1eebb, 1},
+ {0x20000, 0x2a6d6, 1},
+ {0x2a700, 0x2b734, 1},
+ {0x2b740, 0x2b81d, 1},
+ {0x2f800, 0x2fa1d, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Lt = &RangeTable{
+ R16: []Range16{
+ {0x01c5, 0x01cb, 3},
+ {0x01f2, 0x1f88, 7574},
+ {0x1f89, 0x1f8f, 1},
+ {0x1f98, 0x1f9f, 1},
+ {0x1fa8, 0x1faf, 1},
+ {0x1fbc, 0x1fcc, 16},
+ {0x1ffc, 0x1ffc, 1},
+ },
+}
+
+var _Lu = &RangeTable{
+ R16: []Range16{
+ {0x0041, 0x005a, 1},
+ {0x00c0, 0x00d6, 1},
+ {0x00d8, 0x00de, 1},
+ {0x0100, 0x0136, 2},
+ {0x0139, 0x0147, 2},
+ {0x014a, 0x0178, 2},
+ {0x0179, 0x017d, 2},
+ {0x0181, 0x0182, 1},
+ {0x0184, 0x0186, 2},
+ {0x0187, 0x0189, 2},
+ {0x018a, 0x018b, 1},
+ {0x018e, 0x0191, 1},
+ {0x0193, 0x0194, 1},
+ {0x0196, 0x0198, 1},
+ {0x019c, 0x019d, 1},
+ {0x019f, 0x01a0, 1},
+ {0x01a2, 0x01a6, 2},
+ {0x01a7, 0x01a9, 2},
+ {0x01ac, 0x01ae, 2},
+ {0x01af, 0x01b1, 2},
+ {0x01b2, 0x01b3, 1},
+ {0x01b5, 0x01b7, 2},
+ {0x01b8, 0x01bc, 4},
+ {0x01c4, 0x01cd, 3},
+ {0x01cf, 0x01db, 2},
+ {0x01de, 0x01ee, 2},
+ {0x01f1, 0x01f4, 3},
+ {0x01f6, 0x01f8, 1},
+ {0x01fa, 0x0232, 2},
+ {0x023a, 0x023b, 1},
+ {0x023d, 0x023e, 1},
+ {0x0241, 0x0243, 2},
+ {0x0244, 0x0246, 1},
+ {0x0248, 0x024e, 2},
+ {0x0370, 0x0372, 2},
+ {0x0376, 0x037f, 9},
+ {0x0386, 0x0388, 2},
+ {0x0389, 0x038a, 1},
+ {0x038c, 0x038e, 2},
+ {0x038f, 0x0391, 2},
+ {0x0392, 0x03a1, 1},
+ {0x03a3, 0x03ab, 1},
+ {0x03cf, 0x03d2, 3},
+ {0x03d3, 0x03d4, 1},
+ {0x03d8, 0x03ee, 2},
+ {0x03f4, 0x03f7, 3},
+ {0x03f9, 0x03fa, 1},
+ {0x03fd, 0x042f, 1},
+ {0x0460, 0x0480, 2},
+ {0x048a, 0x04c0, 2},
+ {0x04c1, 0x04cd, 2},
+ {0x04d0, 0x052e, 2},
+ {0x0531, 0x0556, 1},
+ {0x10a0, 0x10c5, 1},
+ {0x10c7, 0x10cd, 6},
+ {0x1e00, 0x1e94, 2},
+ {0x1e9e, 0x1efe, 2},
+ {0x1f08, 0x1f0f, 1},
+ {0x1f18, 0x1f1d, 1},
+ {0x1f28, 0x1f2f, 1},
+ {0x1f38, 0x1f3f, 1},
+ {0x1f48, 0x1f4d, 1},
+ {0x1f59, 0x1f5f, 2},
+ {0x1f68, 0x1f6f, 1},
+ {0x1fb8, 0x1fbb, 1},
+ {0x1fc8, 0x1fcb, 1},
+ {0x1fd8, 0x1fdb, 1},
+ {0x1fe8, 0x1fec, 1},
+ {0x1ff8, 0x1ffb, 1},
+ {0x2102, 0x2107, 5},
+ {0x210b, 0x210d, 1},
+ {0x2110, 0x2112, 1},
+ {0x2115, 0x2119, 4},
+ {0x211a, 0x211d, 1},
+ {0x2124, 0x212a, 2},
+ {0x212b, 0x212d, 1},
+ {0x2130, 0x2133, 1},
+ {0x213e, 0x213f, 1},
+ {0x2145, 0x2183, 62},
+ {0x2c00, 0x2c2e, 1},
+ {0x2c60, 0x2c62, 2},
+ {0x2c63, 0x2c64, 1},
+ {0x2c67, 0x2c6d, 2},
+ {0x2c6e, 0x2c70, 1},
+ {0x2c72, 0x2c75, 3},
+ {0x2c7e, 0x2c80, 1},
+ {0x2c82, 0x2ce2, 2},
+ {0x2ceb, 0x2ced, 2},
+ {0x2cf2, 0xa640, 31054},
+ {0xa642, 0xa66c, 2},
+ {0xa680, 0xa69a, 2},
+ {0xa722, 0xa72e, 2},
+ {0xa732, 0xa76e, 2},
+ {0xa779, 0xa77d, 2},
+ {0xa77e, 0xa786, 2},
+ {0xa78b, 0xa78d, 2},
+ {0xa790, 0xa792, 2},
+ {0xa796, 0xa7aa, 2},
+ {0xa7ab, 0xa7ad, 1},
+ {0xa7b0, 0xa7b1, 1},
+ {0xff21, 0xff3a, 1},
+ },
+ R32: []Range32{
+ {0x10400, 0x10427, 1},
+ {0x118a0, 0x118bf, 1},
+ {0x1d400, 0x1d419, 1},
+ {0x1d434, 0x1d44d, 1},
+ {0x1d468, 0x1d481, 1},
+ {0x1d49c, 0x1d49e, 2},
+ {0x1d49f, 0x1d4a5, 3},
+ {0x1d4a6, 0x1d4a9, 3},
+ {0x1d4aa, 0x1d4ac, 1},
+ {0x1d4ae, 0x1d4b5, 1},
+ {0x1d4d0, 0x1d4e9, 1},
+ {0x1d504, 0x1d505, 1},
+ {0x1d507, 0x1d50a, 1},
+ {0x1d50d, 0x1d514, 1},
+ {0x1d516, 0x1d51c, 1},
+ {0x1d538, 0x1d539, 1},
+ {0x1d53b, 0x1d53e, 1},
+ {0x1d540, 0x1d544, 1},
+ {0x1d546, 0x1d54a, 4},
+ {0x1d54b, 0x1d550, 1},
+ {0x1d56c, 0x1d585, 1},
+ {0x1d5a0, 0x1d5b9, 1},
+ {0x1d5d4, 0x1d5ed, 1},
+ {0x1d608, 0x1d621, 1},
+ {0x1d63c, 0x1d655, 1},
+ {0x1d670, 0x1d689, 1},
+ {0x1d6a8, 0x1d6c0, 1},
+ {0x1d6e2, 0x1d6fa, 1},
+ {0x1d71c, 0x1d734, 1},
+ {0x1d756, 0x1d76e, 1},
+ {0x1d790, 0x1d7a8, 1},
+ {0x1d7ca, 0x1d7ca, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _M = &RangeTable{
+ R16: []Range16{
+ {0x0300, 0x036f, 1},
+ {0x0483, 0x0489, 1},
+ {0x0591, 0x05bd, 1},
+ {0x05bf, 0x05c1, 2},
+ {0x05c2, 0x05c4, 2},
+ {0x05c5, 0x05c7, 2},
+ {0x0610, 0x061a, 1},
+ {0x064b, 0x065f, 1},
+ {0x0670, 0x06d6, 102},
+ {0x06d7, 0x06dc, 1},
+ {0x06df, 0x06e4, 1},
+ {0x06e7, 0x06e8, 1},
+ {0x06ea, 0x06ed, 1},
+ {0x0711, 0x0730, 31},
+ {0x0731, 0x074a, 1},
+ {0x07a6, 0x07b0, 1},
+ {0x07eb, 0x07f3, 1},
+ {0x0816, 0x0819, 1},
+ {0x081b, 0x0823, 1},
+ {0x0825, 0x0827, 1},
+ {0x0829, 0x082d, 1},
+ {0x0859, 0x085b, 1},
+ {0x08e4, 0x0903, 1},
+ {0x093a, 0x093c, 1},
+ {0x093e, 0x094f, 1},
+ {0x0951, 0x0957, 1},
+ {0x0962, 0x0963, 1},
+ {0x0981, 0x0983, 1},
+ {0x09bc, 0x09be, 2},
+ {0x09bf, 0x09c4, 1},
+ {0x09c7, 0x09c8, 1},
+ {0x09cb, 0x09cd, 1},
+ {0x09d7, 0x09e2, 11},
+ {0x09e3, 0x0a01, 30},
+ {0x0a02, 0x0a03, 1},
+ {0x0a3c, 0x0a3e, 2},
+ {0x0a3f, 0x0a42, 1},
+ {0x0a47, 0x0a48, 1},
+ {0x0a4b, 0x0a4d, 1},
+ {0x0a51, 0x0a70, 31},
+ {0x0a71, 0x0a75, 4},
+ {0x0a81, 0x0a83, 1},
+ {0x0abc, 0x0abe, 2},
+ {0x0abf, 0x0ac5, 1},
+ {0x0ac7, 0x0ac9, 1},
+ {0x0acb, 0x0acd, 1},
+ {0x0ae2, 0x0ae3, 1},
+ {0x0b01, 0x0b03, 1},
+ {0x0b3c, 0x0b3e, 2},
+ {0x0b3f, 0x0b44, 1},
+ {0x0b47, 0x0b48, 1},
+ {0x0b4b, 0x0b4d, 1},
+ {0x0b56, 0x0b57, 1},
+ {0x0b62, 0x0b63, 1},
+ {0x0b82, 0x0bbe, 60},
+ {0x0bbf, 0x0bc2, 1},
+ {0x0bc6, 0x0bc8, 1},
+ {0x0bca, 0x0bcd, 1},
+ {0x0bd7, 0x0c00, 41},
+ {0x0c01, 0x0c03, 1},
+ {0x0c3e, 0x0c44, 1},
+ {0x0c46, 0x0c48, 1},
+ {0x0c4a, 0x0c4d, 1},
+ {0x0c55, 0x0c56, 1},
+ {0x0c62, 0x0c63, 1},
+ {0x0c81, 0x0c83, 1},
+ {0x0cbc, 0x0cbe, 2},
+ {0x0cbf, 0x0cc4, 1},
+ {0x0cc6, 0x0cc8, 1},
+ {0x0cca, 0x0ccd, 1},
+ {0x0cd5, 0x0cd6, 1},
+ {0x0ce2, 0x0ce3, 1},
+ {0x0d01, 0x0d03, 1},
+ {0x0d3e, 0x0d44, 1},
+ {0x0d46, 0x0d48, 1},
+ {0x0d4a, 0x0d4d, 1},
+ {0x0d57, 0x0d62, 11},
+ {0x0d63, 0x0d82, 31},
+ {0x0d83, 0x0dca, 71},
+ {0x0dcf, 0x0dd4, 1},
+ {0x0dd6, 0x0dd8, 2},
+ {0x0dd9, 0x0ddf, 1},
+ {0x0df2, 0x0df3, 1},
+ {0x0e31, 0x0e34, 3},
+ {0x0e35, 0x0e3a, 1},
+ {0x0e47, 0x0e4e, 1},
+ {0x0eb1, 0x0eb4, 3},
+ {0x0eb5, 0x0eb9, 1},
+ {0x0ebb, 0x0ebc, 1},
+ {0x0ec8, 0x0ecd, 1},
+ {0x0f18, 0x0f19, 1},
+ {0x0f35, 0x0f39, 2},
+ {0x0f3e, 0x0f3f, 1},
+ {0x0f71, 0x0f84, 1},
+ {0x0f86, 0x0f87, 1},
+ {0x0f8d, 0x0f97, 1},
+ {0x0f99, 0x0fbc, 1},
+ {0x0fc6, 0x102b, 101},
+ {0x102c, 0x103e, 1},
+ {0x1056, 0x1059, 1},
+ {0x105e, 0x1060, 1},
+ {0x1062, 0x1064, 1},
+ {0x1067, 0x106d, 1},
+ {0x1071, 0x1074, 1},
+ {0x1082, 0x108d, 1},
+ {0x108f, 0x109a, 11},
+ {0x109b, 0x109d, 1},
+ {0x135d, 0x135f, 1},
+ {0x1712, 0x1714, 1},
+ {0x1732, 0x1734, 1},
+ {0x1752, 0x1753, 1},
+ {0x1772, 0x1773, 1},
+ {0x17b4, 0x17d3, 1},
+ {0x17dd, 0x180b, 46},
+ {0x180c, 0x180d, 1},
+ {0x18a9, 0x1920, 119},
+ {0x1921, 0x192b, 1},
+ {0x1930, 0x193b, 1},
+ {0x19b0, 0x19c0, 1},
+ {0x19c8, 0x19c9, 1},
+ {0x1a17, 0x1a1b, 1},
+ {0x1a55, 0x1a5e, 1},
+ {0x1a60, 0x1a7c, 1},
+ {0x1a7f, 0x1ab0, 49},
+ {0x1ab1, 0x1abe, 1},
+ {0x1b00, 0x1b04, 1},
+ {0x1b34, 0x1b44, 1},
+ {0x1b6b, 0x1b73, 1},
+ {0x1b80, 0x1b82, 1},
+ {0x1ba1, 0x1bad, 1},
+ {0x1be6, 0x1bf3, 1},
+ {0x1c24, 0x1c37, 1},
+ {0x1cd0, 0x1cd2, 1},
+ {0x1cd4, 0x1ce8, 1},
+ {0x1ced, 0x1cf2, 5},
+ {0x1cf3, 0x1cf4, 1},
+ {0x1cf8, 0x1cf9, 1},
+ {0x1dc0, 0x1df5, 1},
+ {0x1dfc, 0x1dff, 1},
+ {0x20d0, 0x20f0, 1},
+ {0x2cef, 0x2cf1, 1},
+ {0x2d7f, 0x2de0, 97},
+ {0x2de1, 0x2dff, 1},
+ {0x302a, 0x302f, 1},
+ {0x3099, 0x309a, 1},
+ {0xa66f, 0xa672, 1},
+ {0xa674, 0xa67d, 1},
+ {0xa69f, 0xa6f0, 81},
+ {0xa6f1, 0xa802, 273},
+ {0xa806, 0xa80b, 5},
+ {0xa823, 0xa827, 1},
+ {0xa880, 0xa881, 1},
+ {0xa8b4, 0xa8c4, 1},
+ {0xa8e0, 0xa8f1, 1},
+ {0xa926, 0xa92d, 1},
+ {0xa947, 0xa953, 1},
+ {0xa980, 0xa983, 1},
+ {0xa9b3, 0xa9c0, 1},
+ {0xa9e5, 0xaa29, 68},
+ {0xaa2a, 0xaa36, 1},
+ {0xaa43, 0xaa4c, 9},
+ {0xaa4d, 0xaa7b, 46},
+ {0xaa7c, 0xaa7d, 1},
+ {0xaab0, 0xaab2, 2},
+ {0xaab3, 0xaab4, 1},
+ {0xaab7, 0xaab8, 1},
+ {0xaabe, 0xaabf, 1},
+ {0xaac1, 0xaaeb, 42},
+ {0xaaec, 0xaaef, 1},
+ {0xaaf5, 0xaaf6, 1},
+ {0xabe3, 0xabea, 1},
+ {0xabec, 0xabed, 1},
+ {0xfb1e, 0xfe00, 738},
+ {0xfe01, 0xfe0f, 1},
+ {0xfe20, 0xfe2d, 1},
+ },
+ R32: []Range32{
+ {0x101fd, 0x102e0, 227},
+ {0x10376, 0x1037a, 1},
+ {0x10a01, 0x10a03, 1},
+ {0x10a05, 0x10a06, 1},
+ {0x10a0c, 0x10a0f, 1},
+ {0x10a38, 0x10a3a, 1},
+ {0x10a3f, 0x10ae5, 166},
+ {0x10ae6, 0x11000, 1306},
+ {0x11001, 0x11002, 1},
+ {0x11038, 0x11046, 1},
+ {0x1107f, 0x11082, 1},
+ {0x110b0, 0x110ba, 1},
+ {0x11100, 0x11102, 1},
+ {0x11127, 0x11134, 1},
+ {0x11173, 0x11180, 13},
+ {0x11181, 0x11182, 1},
+ {0x111b3, 0x111c0, 1},
+ {0x1122c, 0x11237, 1},
+ {0x112df, 0x112ea, 1},
+ {0x11301, 0x11303, 1},
+ {0x1133c, 0x1133e, 2},
+ {0x1133f, 0x11344, 1},
+ {0x11347, 0x11348, 1},
+ {0x1134b, 0x1134d, 1},
+ {0x11357, 0x11362, 11},
+ {0x11363, 0x11366, 3},
+ {0x11367, 0x1136c, 1},
+ {0x11370, 0x11374, 1},
+ {0x114b0, 0x114c3, 1},
+ {0x115af, 0x115b5, 1},
+ {0x115b8, 0x115c0, 1},
+ {0x11630, 0x11640, 1},
+ {0x116ab, 0x116b7, 1},
+ {0x16af0, 0x16af4, 1},
+ {0x16b30, 0x16b36, 1},
+ {0x16f51, 0x16f7e, 1},
+ {0x16f8f, 0x16f92, 1},
+ {0x1bc9d, 0x1bc9e, 1},
+ {0x1d165, 0x1d169, 1},
+ {0x1d16d, 0x1d172, 1},
+ {0x1d17b, 0x1d182, 1},
+ {0x1d185, 0x1d18b, 1},
+ {0x1d1aa, 0x1d1ad, 1},
+ {0x1d242, 0x1d244, 1},
+ {0x1e8d0, 0x1e8d6, 1},
+ {0xe0100, 0xe01ef, 1},
+ },
+}
+
+var _Mc = &RangeTable{
+ R16: []Range16{
+ {0x0903, 0x093b, 56},
+ {0x093e, 0x0940, 1},
+ {0x0949, 0x094c, 1},
+ {0x094e, 0x094f, 1},
+ {0x0982, 0x0983, 1},
+ {0x09be, 0x09c0, 1},
+ {0x09c7, 0x09c8, 1},
+ {0x09cb, 0x09cc, 1},
+ {0x09d7, 0x0a03, 44},
+ {0x0a3e, 0x0a40, 1},
+ {0x0a83, 0x0abe, 59},
+ {0x0abf, 0x0ac0, 1},
+ {0x0ac9, 0x0acb, 2},
+ {0x0acc, 0x0b02, 54},
+ {0x0b03, 0x0b3e, 59},
+ {0x0b40, 0x0b47, 7},
+ {0x0b48, 0x0b4b, 3},
+ {0x0b4c, 0x0b57, 11},
+ {0x0bbe, 0x0bbf, 1},
+ {0x0bc1, 0x0bc2, 1},
+ {0x0bc6, 0x0bc8, 1},
+ {0x0bca, 0x0bcc, 1},
+ {0x0bd7, 0x0c01, 42},
+ {0x0c02, 0x0c03, 1},
+ {0x0c41, 0x0c44, 1},
+ {0x0c82, 0x0c83, 1},
+ {0x0cbe, 0x0cc0, 2},
+ {0x0cc1, 0x0cc4, 1},
+ {0x0cc7, 0x0cc8, 1},
+ {0x0cca, 0x0ccb, 1},
+ {0x0cd5, 0x0cd6, 1},
+ {0x0d02, 0x0d03, 1},
+ {0x0d3e, 0x0d40, 1},
+ {0x0d46, 0x0d48, 1},
+ {0x0d4a, 0x0d4c, 1},
+ {0x0d57, 0x0d82, 43},
+ {0x0d83, 0x0dcf, 76},
+ {0x0dd0, 0x0dd1, 1},
+ {0x0dd8, 0x0ddf, 1},
+ {0x0df2, 0x0df3, 1},
+ {0x0f3e, 0x0f3f, 1},
+ {0x0f7f, 0x102b, 172},
+ {0x102c, 0x1031, 5},
+ {0x1038, 0x103b, 3},
+ {0x103c, 0x1056, 26},
+ {0x1057, 0x1062, 11},
+ {0x1063, 0x1064, 1},
+ {0x1067, 0x106d, 1},
+ {0x1083, 0x1084, 1},
+ {0x1087, 0x108c, 1},
+ {0x108f, 0x109a, 11},
+ {0x109b, 0x109c, 1},
+ {0x17b6, 0x17be, 8},
+ {0x17bf, 0x17c5, 1},
+ {0x17c7, 0x17c8, 1},
+ {0x1923, 0x1926, 1},
+ {0x1929, 0x192b, 1},
+ {0x1930, 0x1931, 1},
+ {0x1933, 0x1938, 1},
+ {0x19b0, 0x19c0, 1},
+ {0x19c8, 0x19c9, 1},
+ {0x1a19, 0x1a1a, 1},
+ {0x1a55, 0x1a57, 2},
+ {0x1a61, 0x1a63, 2},
+ {0x1a64, 0x1a6d, 9},
+ {0x1a6e, 0x1a72, 1},
+ {0x1b04, 0x1b35, 49},
+ {0x1b3b, 0x1b3d, 2},
+ {0x1b3e, 0x1b41, 1},
+ {0x1b43, 0x1b44, 1},
+ {0x1b82, 0x1ba1, 31},
+ {0x1ba6, 0x1ba7, 1},
+ {0x1baa, 0x1be7, 61},
+ {0x1bea, 0x1bec, 1},
+ {0x1bee, 0x1bf2, 4},
+ {0x1bf3, 0x1c24, 49},
+ {0x1c25, 0x1c2b, 1},
+ {0x1c34, 0x1c35, 1},
+ {0x1ce1, 0x1cf2, 17},
+ {0x1cf3, 0x302e, 4923},
+ {0x302f, 0xa823, 30708},
+ {0xa824, 0xa827, 3},
+ {0xa880, 0xa881, 1},
+ {0xa8b4, 0xa8c3, 1},
+ {0xa952, 0xa953, 1},
+ {0xa983, 0xa9b4, 49},
+ {0xa9b5, 0xa9ba, 5},
+ {0xa9bb, 0xa9bd, 2},
+ {0xa9be, 0xa9c0, 1},
+ {0xaa2f, 0xaa30, 1},
+ {0xaa33, 0xaa34, 1},
+ {0xaa4d, 0xaa7b, 46},
+ {0xaa7d, 0xaaeb, 110},
+ {0xaaee, 0xaaef, 1},
+ {0xaaf5, 0xabe3, 238},
+ {0xabe4, 0xabe6, 2},
+ {0xabe7, 0xabe9, 2},
+ {0xabea, 0xabec, 2},
+ },
+ R32: []Range32{
+ {0x11000, 0x11002, 2},
+ {0x11082, 0x110b0, 46},
+ {0x110b1, 0x110b2, 1},
+ {0x110b7, 0x110b8, 1},
+ {0x1112c, 0x11182, 86},
+ {0x111b3, 0x111b5, 1},
+ {0x111bf, 0x111c0, 1},
+ {0x1122c, 0x1122e, 1},
+ {0x11232, 0x11233, 1},
+ {0x11235, 0x112e0, 171},
+ {0x112e1, 0x112e2, 1},
+ {0x11302, 0x11303, 1},
+ {0x1133e, 0x1133f, 1},
+ {0x11341, 0x11344, 1},
+ {0x11347, 0x11348, 1},
+ {0x1134b, 0x1134d, 1},
+ {0x11357, 0x11362, 11},
+ {0x11363, 0x114b0, 333},
+ {0x114b1, 0x114b2, 1},
+ {0x114b9, 0x114bb, 2},
+ {0x114bc, 0x114be, 1},
+ {0x114c1, 0x115af, 238},
+ {0x115b0, 0x115b1, 1},
+ {0x115b8, 0x115bb, 1},
+ {0x115be, 0x11630, 114},
+ {0x11631, 0x11632, 1},
+ {0x1163b, 0x1163c, 1},
+ {0x1163e, 0x116ac, 110},
+ {0x116ae, 0x116af, 1},
+ {0x116b6, 0x16f51, 22683},
+ {0x16f52, 0x16f7e, 1},
+ {0x1d165, 0x1d166, 1},
+ {0x1d16d, 0x1d172, 1},
+ },
+}
+
+var _Me = &RangeTable{
+ R16: []Range16{
+ {0x0488, 0x0489, 1},
+ {0x1abe, 0x20dd, 1567},
+ {0x20de, 0x20e0, 1},
+ {0x20e2, 0x20e4, 1},
+ {0xa670, 0xa672, 1},
+ },
+}
+
+var _Mn = &RangeTable{
+ R16: []Range16{
+ {0x0300, 0x036f, 1},
+ {0x0483, 0x0487, 1},
+ {0x0591, 0x05bd, 1},
+ {0x05bf, 0x05c1, 2},
+ {0x05c2, 0x05c4, 2},
+ {0x05c5, 0x05c7, 2},
+ {0x0610, 0x061a, 1},
+ {0x064b, 0x065f, 1},
+ {0x0670, 0x06d6, 102},
+ {0x06d7, 0x06dc, 1},
+ {0x06df, 0x06e4, 1},
+ {0x06e7, 0x06e8, 1},
+ {0x06ea, 0x06ed, 1},
+ {0x0711, 0x0730, 31},
+ {0x0731, 0x074a, 1},
+ {0x07a6, 0x07b0, 1},
+ {0x07eb, 0x07f3, 1},
+ {0x0816, 0x0819, 1},
+ {0x081b, 0x0823, 1},
+ {0x0825, 0x0827, 1},
+ {0x0829, 0x082d, 1},
+ {0x0859, 0x085b, 1},
+ {0x08e4, 0x0902, 1},
+ {0x093a, 0x093c, 2},
+ {0x0941, 0x0948, 1},
+ {0x094d, 0x0951, 4},
+ {0x0952, 0x0957, 1},
+ {0x0962, 0x0963, 1},
+ {0x0981, 0x09bc, 59},
+ {0x09c1, 0x09c4, 1},
+ {0x09cd, 0x09e2, 21},
+ {0x09e3, 0x0a01, 30},
+ {0x0a02, 0x0a3c, 58},
+ {0x0a41, 0x0a42, 1},
+ {0x0a47, 0x0a48, 1},
+ {0x0a4b, 0x0a4d, 1},
+ {0x0a51, 0x0a70, 31},
+ {0x0a71, 0x0a75, 4},
+ {0x0a81, 0x0a82, 1},
+ {0x0abc, 0x0ac1, 5},
+ {0x0ac2, 0x0ac5, 1},
+ {0x0ac7, 0x0ac8, 1},
+ {0x0acd, 0x0ae2, 21},
+ {0x0ae3, 0x0b01, 30},
+ {0x0b3c, 0x0b3f, 3},
+ {0x0b41, 0x0b44, 1},
+ {0x0b4d, 0x0b56, 9},
+ {0x0b62, 0x0b63, 1},
+ {0x0b82, 0x0bc0, 62},
+ {0x0bcd, 0x0c00, 51},
+ {0x0c3e, 0x0c40, 1},
+ {0x0c46, 0x0c48, 1},
+ {0x0c4a, 0x0c4d, 1},
+ {0x0c55, 0x0c56, 1},
+ {0x0c62, 0x0c63, 1},
+ {0x0c81, 0x0cbc, 59},
+ {0x0cbf, 0x0cc6, 7},
+ {0x0ccc, 0x0ccd, 1},
+ {0x0ce2, 0x0ce3, 1},
+ {0x0d01, 0x0d41, 64},
+ {0x0d42, 0x0d44, 1},
+ {0x0d4d, 0x0d62, 21},
+ {0x0d63, 0x0dca, 103},
+ {0x0dd2, 0x0dd4, 1},
+ {0x0dd6, 0x0e31, 91},
+ {0x0e34, 0x0e3a, 1},
+ {0x0e47, 0x0e4e, 1},
+ {0x0eb1, 0x0eb4, 3},
+ {0x0eb5, 0x0eb9, 1},
+ {0x0ebb, 0x0ebc, 1},
+ {0x0ec8, 0x0ecd, 1},
+ {0x0f18, 0x0f19, 1},
+ {0x0f35, 0x0f39, 2},
+ {0x0f71, 0x0f7e, 1},
+ {0x0f80, 0x0f84, 1},
+ {0x0f86, 0x0f87, 1},
+ {0x0f8d, 0x0f97, 1},
+ {0x0f99, 0x0fbc, 1},
+ {0x0fc6, 0x102d, 103},
+ {0x102e, 0x1030, 1},
+ {0x1032, 0x1037, 1},
+ {0x1039, 0x103a, 1},
+ {0x103d, 0x103e, 1},
+ {0x1058, 0x1059, 1},
+ {0x105e, 0x1060, 1},
+ {0x1071, 0x1074, 1},
+ {0x1082, 0x1085, 3},
+ {0x1086, 0x108d, 7},
+ {0x109d, 0x135d, 704},
+ {0x135e, 0x135f, 1},
+ {0x1712, 0x1714, 1},
+ {0x1732, 0x1734, 1},
+ {0x1752, 0x1753, 1},
+ {0x1772, 0x1773, 1},
+ {0x17b4, 0x17b5, 1},
+ {0x17b7, 0x17bd, 1},
+ {0x17c6, 0x17c9, 3},
+ {0x17ca, 0x17d3, 1},
+ {0x17dd, 0x180b, 46},
+ {0x180c, 0x180d, 1},
+ {0x18a9, 0x1920, 119},
+ {0x1921, 0x1922, 1},
+ {0x1927, 0x1928, 1},
+ {0x1932, 0x1939, 7},
+ {0x193a, 0x193b, 1},
+ {0x1a17, 0x1a18, 1},
+ {0x1a1b, 0x1a56, 59},
+ {0x1a58, 0x1a5e, 1},
+ {0x1a60, 0x1a62, 2},
+ {0x1a65, 0x1a6c, 1},
+ {0x1a73, 0x1a7c, 1},
+ {0x1a7f, 0x1ab0, 49},
+ {0x1ab1, 0x1abd, 1},
+ {0x1b00, 0x1b03, 1},
+ {0x1b34, 0x1b36, 2},
+ {0x1b37, 0x1b3a, 1},
+ {0x1b3c, 0x1b42, 6},
+ {0x1b6b, 0x1b73, 1},
+ {0x1b80, 0x1b81, 1},
+ {0x1ba2, 0x1ba5, 1},
+ {0x1ba8, 0x1ba9, 1},
+ {0x1bab, 0x1bad, 1},
+ {0x1be6, 0x1be8, 2},
+ {0x1be9, 0x1bed, 4},
+ {0x1bef, 0x1bf1, 1},
+ {0x1c2c, 0x1c33, 1},
+ {0x1c36, 0x1c37, 1},
+ {0x1cd0, 0x1cd2, 1},
+ {0x1cd4, 0x1ce0, 1},
+ {0x1ce2, 0x1ce8, 1},
+ {0x1ced, 0x1cf4, 7},
+ {0x1cf8, 0x1cf9, 1},
+ {0x1dc0, 0x1df5, 1},
+ {0x1dfc, 0x1dff, 1},
+ {0x20d0, 0x20dc, 1},
+ {0x20e1, 0x20e5, 4},
+ {0x20e6, 0x20f0, 1},
+ {0x2cef, 0x2cf1, 1},
+ {0x2d7f, 0x2de0, 97},
+ {0x2de1, 0x2dff, 1},
+ {0x302a, 0x302d, 1},
+ {0x3099, 0x309a, 1},
+ {0xa66f, 0xa674, 5},
+ {0xa675, 0xa67d, 1},
+ {0xa69f, 0xa6f0, 81},
+ {0xa6f1, 0xa802, 273},
+ {0xa806, 0xa80b, 5},
+ {0xa825, 0xa826, 1},
+ {0xa8c4, 0xa8e0, 28},
+ {0xa8e1, 0xa8f1, 1},
+ {0xa926, 0xa92d, 1},
+ {0xa947, 0xa951, 1},
+ {0xa980, 0xa982, 1},
+ {0xa9b3, 0xa9b6, 3},
+ {0xa9b7, 0xa9b9, 1},
+ {0xa9bc, 0xa9e5, 41},
+ {0xaa29, 0xaa2e, 1},
+ {0xaa31, 0xaa32, 1},
+ {0xaa35, 0xaa36, 1},
+ {0xaa43, 0xaa4c, 9},
+ {0xaa7c, 0xaab0, 52},
+ {0xaab2, 0xaab4, 1},
+ {0xaab7, 0xaab8, 1},
+ {0xaabe, 0xaabf, 1},
+ {0xaac1, 0xaaec, 43},
+ {0xaaed, 0xaaf6, 9},
+ {0xabe5, 0xabe8, 3},
+ {0xabed, 0xfb1e, 20273},
+ {0xfe00, 0xfe0f, 1},
+ {0xfe20, 0xfe2d, 1},
+ },
+ R32: []Range32{
+ {0x101fd, 0x102e0, 227},
+ {0x10376, 0x1037a, 1},
+ {0x10a01, 0x10a03, 1},
+ {0x10a05, 0x10a06, 1},
+ {0x10a0c, 0x10a0f, 1},
+ {0x10a38, 0x10a3a, 1},
+ {0x10a3f, 0x10ae5, 166},
+ {0x10ae6, 0x11001, 1307},
+ {0x11038, 0x11046, 1},
+ {0x1107f, 0x11081, 1},
+ {0x110b3, 0x110b6, 1},
+ {0x110b9, 0x110ba, 1},
+ {0x11100, 0x11102, 1},
+ {0x11127, 0x1112b, 1},
+ {0x1112d, 0x11134, 1},
+ {0x11173, 0x11180, 13},
+ {0x11181, 0x111b6, 53},
+ {0x111b7, 0x111be, 1},
+ {0x1122f, 0x11231, 1},
+ {0x11234, 0x11236, 2},
+ {0x11237, 0x112df, 168},
+ {0x112e3, 0x112ea, 1},
+ {0x11301, 0x1133c, 59},
+ {0x11340, 0x11366, 38},
+ {0x11367, 0x1136c, 1},
+ {0x11370, 0x11374, 1},
+ {0x114b3, 0x114b8, 1},
+ {0x114ba, 0x114bf, 5},
+ {0x114c0, 0x114c2, 2},
+ {0x114c3, 0x115b2, 239},
+ {0x115b3, 0x115b5, 1},
+ {0x115bc, 0x115bd, 1},
+ {0x115bf, 0x115c0, 1},
+ {0x11633, 0x1163a, 1},
+ {0x1163d, 0x1163f, 2},
+ {0x11640, 0x116ab, 107},
+ {0x116ad, 0x116b0, 3},
+ {0x116b1, 0x116b5, 1},
+ {0x116b7, 0x16af0, 21561},
+ {0x16af1, 0x16af4, 1},
+ {0x16b30, 0x16b36, 1},
+ {0x16f8f, 0x16f92, 1},
+ {0x1bc9d, 0x1bc9e, 1},
+ {0x1d167, 0x1d169, 1},
+ {0x1d17b, 0x1d182, 1},
+ {0x1d185, 0x1d18b, 1},
+ {0x1d1aa, 0x1d1ad, 1},
+ {0x1d242, 0x1d244, 1},
+ {0x1e8d0, 0x1e8d6, 1},
+ {0xe0100, 0xe01ef, 1},
+ },
+}
+
+var _N = &RangeTable{
+ R16: []Range16{
+ {0x0030, 0x0039, 1},
+ {0x00b2, 0x00b3, 1},
+ {0x00b9, 0x00bc, 3},
+ {0x00bd, 0x00be, 1},
+ {0x0660, 0x0669, 1},
+ {0x06f0, 0x06f9, 1},
+ {0x07c0, 0x07c9, 1},
+ {0x0966, 0x096f, 1},
+ {0x09e6, 0x09ef, 1},
+ {0x09f4, 0x09f9, 1},
+ {0x0a66, 0x0a6f, 1},
+ {0x0ae6, 0x0aef, 1},
+ {0x0b66, 0x0b6f, 1},
+ {0x0b72, 0x0b77, 1},
+ {0x0be6, 0x0bf2, 1},
+ {0x0c66, 0x0c6f, 1},
+ {0x0c78, 0x0c7e, 1},
+ {0x0ce6, 0x0cef, 1},
+ {0x0d66, 0x0d75, 1},
+ {0x0de6, 0x0def, 1},
+ {0x0e50, 0x0e59, 1},
+ {0x0ed0, 0x0ed9, 1},
+ {0x0f20, 0x0f33, 1},
+ {0x1040, 0x1049, 1},
+ {0x1090, 0x1099, 1},
+ {0x1369, 0x137c, 1},
+ {0x16ee, 0x16f0, 1},
+ {0x17e0, 0x17e9, 1},
+ {0x17f0, 0x17f9, 1},
+ {0x1810, 0x1819, 1},
+ {0x1946, 0x194f, 1},
+ {0x19d0, 0x19da, 1},
+ {0x1a80, 0x1a89, 1},
+ {0x1a90, 0x1a99, 1},
+ {0x1b50, 0x1b59, 1},
+ {0x1bb0, 0x1bb9, 1},
+ {0x1c40, 0x1c49, 1},
+ {0x1c50, 0x1c59, 1},
+ {0x2070, 0x2074, 4},
+ {0x2075, 0x2079, 1},
+ {0x2080, 0x2089, 1},
+ {0x2150, 0x2182, 1},
+ {0x2185, 0x2189, 1},
+ {0x2460, 0x249b, 1},
+ {0x24ea, 0x24ff, 1},
+ {0x2776, 0x2793, 1},
+ {0x2cfd, 0x3007, 778},
+ {0x3021, 0x3029, 1},
+ {0x3038, 0x303a, 1},
+ {0x3192, 0x3195, 1},
+ {0x3220, 0x3229, 1},
+ {0x3248, 0x324f, 1},
+ {0x3251, 0x325f, 1},
+ {0x3280, 0x3289, 1},
+ {0x32b1, 0x32bf, 1},
+ {0xa620, 0xa629, 1},
+ {0xa6e6, 0xa6ef, 1},
+ {0xa830, 0xa835, 1},
+ {0xa8d0, 0xa8d9, 1},
+ {0xa900, 0xa909, 1},
+ {0xa9d0, 0xa9d9, 1},
+ {0xa9f0, 0xa9f9, 1},
+ {0xaa50, 0xaa59, 1},
+ {0xabf0, 0xabf9, 1},
+ {0xff10, 0xff19, 1},
+ },
+ R32: []Range32{
+ {0x10107, 0x10133, 1},
+ {0x10140, 0x10178, 1},
+ {0x1018a, 0x1018b, 1},
+ {0x102e1, 0x102fb, 1},
+ {0x10320, 0x10323, 1},
+ {0x10341, 0x1034a, 9},
+ {0x103d1, 0x103d5, 1},
+ {0x104a0, 0x104a9, 1},
+ {0x10858, 0x1085f, 1},
+ {0x10879, 0x1087f, 1},
+ {0x108a7, 0x108af, 1},
+ {0x10916, 0x1091b, 1},
+ {0x10a40, 0x10a47, 1},
+ {0x10a7d, 0x10a7e, 1},
+ {0x10a9d, 0x10a9f, 1},
+ {0x10aeb, 0x10aef, 1},
+ {0x10b58, 0x10b5f, 1},
+ {0x10b78, 0x10b7f, 1},
+ {0x10ba9, 0x10baf, 1},
+ {0x10e60, 0x10e7e, 1},
+ {0x11052, 0x1106f, 1},
+ {0x110f0, 0x110f9, 1},
+ {0x11136, 0x1113f, 1},
+ {0x111d0, 0x111d9, 1},
+ {0x111e1, 0x111f4, 1},
+ {0x112f0, 0x112f9, 1},
+ {0x114d0, 0x114d9, 1},
+ {0x11650, 0x11659, 1},
+ {0x116c0, 0x116c9, 1},
+ {0x118e0, 0x118f2, 1},
+ {0x12400, 0x1246e, 1},
+ {0x16a60, 0x16a69, 1},
+ {0x16b50, 0x16b59, 1},
+ {0x16b5b, 0x16b61, 1},
+ {0x1d360, 0x1d371, 1},
+ {0x1d7ce, 0x1d7ff, 1},
+ {0x1e8c7, 0x1e8cf, 1},
+ {0x1f100, 0x1f10c, 1},
+ },
+ LatinOffset: 4,
+}
+
+var _Nd = &RangeTable{
+ R16: []Range16{
+ {0x0030, 0x0039, 1},
+ {0x0660, 0x0669, 1},
+ {0x06f0, 0x06f9, 1},
+ {0x07c0, 0x07c9, 1},
+ {0x0966, 0x096f, 1},
+ {0x09e6, 0x09ef, 1},
+ {0x0a66, 0x0a6f, 1},
+ {0x0ae6, 0x0aef, 1},
+ {0x0b66, 0x0b6f, 1},
+ {0x0be6, 0x0bef, 1},
+ {0x0c66, 0x0c6f, 1},
+ {0x0ce6, 0x0cef, 1},
+ {0x0d66, 0x0d6f, 1},
+ {0x0de6, 0x0def, 1},
+ {0x0e50, 0x0e59, 1},
+ {0x0ed0, 0x0ed9, 1},
+ {0x0f20, 0x0f29, 1},
+ {0x1040, 0x1049, 1},
+ {0x1090, 0x1099, 1},
+ {0x17e0, 0x17e9, 1},
+ {0x1810, 0x1819, 1},
+ {0x1946, 0x194f, 1},
+ {0x19d0, 0x19d9, 1},
+ {0x1a80, 0x1a89, 1},
+ {0x1a90, 0x1a99, 1},
+ {0x1b50, 0x1b59, 1},
+ {0x1bb0, 0x1bb9, 1},
+ {0x1c40, 0x1c49, 1},
+ {0x1c50, 0x1c59, 1},
+ {0xa620, 0xa629, 1},
+ {0xa8d0, 0xa8d9, 1},
+ {0xa900, 0xa909, 1},
+ {0xa9d0, 0xa9d9, 1},
+ {0xa9f0, 0xa9f9, 1},
+ {0xaa50, 0xaa59, 1},
+ {0xabf0, 0xabf9, 1},
+ {0xff10, 0xff19, 1},
+ },
+ R32: []Range32{
+ {0x104a0, 0x104a9, 1},
+ {0x11066, 0x1106f, 1},
+ {0x110f0, 0x110f9, 1},
+ {0x11136, 0x1113f, 1},
+ {0x111d0, 0x111d9, 1},
+ {0x112f0, 0x112f9, 1},
+ {0x114d0, 0x114d9, 1},
+ {0x11650, 0x11659, 1},
+ {0x116c0, 0x116c9, 1},
+ {0x118e0, 0x118e9, 1},
+ {0x16a60, 0x16a69, 1},
+ {0x16b50, 0x16b59, 1},
+ {0x1d7ce, 0x1d7ff, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Nl = &RangeTable{
+ R16: []Range16{
+ {0x16ee, 0x16f0, 1},
+ {0x2160, 0x2182, 1},
+ {0x2185, 0x2188, 1},
+ {0x3007, 0x3021, 26},
+ {0x3022, 0x3029, 1},
+ {0x3038, 0x303a, 1},
+ {0xa6e6, 0xa6ef, 1},
+ },
+ R32: []Range32{
+ {0x10140, 0x10174, 1},
+ {0x10341, 0x1034a, 9},
+ {0x103d1, 0x103d5, 1},
+ {0x12400, 0x1246e, 1},
+ },
+}
+
+var _No = &RangeTable{
+ R16: []Range16{
+ {0x00b2, 0x00b3, 1},
+ {0x00b9, 0x00bc, 3},
+ {0x00bd, 0x00be, 1},
+ {0x09f4, 0x09f9, 1},
+ {0x0b72, 0x0b77, 1},
+ {0x0bf0, 0x0bf2, 1},
+ {0x0c78, 0x0c7e, 1},
+ {0x0d70, 0x0d75, 1},
+ {0x0f2a, 0x0f33, 1},
+ {0x1369, 0x137c, 1},
+ {0x17f0, 0x17f9, 1},
+ {0x19da, 0x2070, 1686},
+ {0x2074, 0x2079, 1},
+ {0x2080, 0x2089, 1},
+ {0x2150, 0x215f, 1},
+ {0x2189, 0x2460, 727},
+ {0x2461, 0x249b, 1},
+ {0x24ea, 0x24ff, 1},
+ {0x2776, 0x2793, 1},
+ {0x2cfd, 0x3192, 1173},
+ {0x3193, 0x3195, 1},
+ {0x3220, 0x3229, 1},
+ {0x3248, 0x324f, 1},
+ {0x3251, 0x325f, 1},
+ {0x3280, 0x3289, 1},
+ {0x32b1, 0x32bf, 1},
+ {0xa830, 0xa835, 1},
+ },
+ R32: []Range32{
+ {0x10107, 0x10133, 1},
+ {0x10175, 0x10178, 1},
+ {0x1018a, 0x1018b, 1},
+ {0x102e1, 0x102fb, 1},
+ {0x10320, 0x10323, 1},
+ {0x10858, 0x1085f, 1},
+ {0x10879, 0x1087f, 1},
+ {0x108a7, 0x108af, 1},
+ {0x10916, 0x1091b, 1},
+ {0x10a40, 0x10a47, 1},
+ {0x10a7d, 0x10a7e, 1},
+ {0x10a9d, 0x10a9f, 1},
+ {0x10aeb, 0x10aef, 1},
+ {0x10b58, 0x10b5f, 1},
+ {0x10b78, 0x10b7f, 1},
+ {0x10ba9, 0x10baf, 1},
+ {0x10e60, 0x10e7e, 1},
+ {0x11052, 0x11065, 1},
+ {0x111e1, 0x111f4, 1},
+ {0x118ea, 0x118f2, 1},
+ {0x16b5b, 0x16b61, 1},
+ {0x1d360, 0x1d371, 1},
+ {0x1e8c7, 0x1e8cf, 1},
+ {0x1f100, 0x1f10c, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _P = &RangeTable{
+ R16: []Range16{
+ {0x0021, 0x0023, 1},
+ {0x0025, 0x002a, 1},
+ {0x002c, 0x002f, 1},
+ {0x003a, 0x003b, 1},
+ {0x003f, 0x0040, 1},
+ {0x005b, 0x005d, 1},
+ {0x005f, 0x007b, 28},
+ {0x007d, 0x00a1, 36},
+ {0x00a7, 0x00ab, 4},
+ {0x00b6, 0x00b7, 1},
+ {0x00bb, 0x00bf, 4},
+ {0x037e, 0x0387, 9},
+ {0x055a, 0x055f, 1},
+ {0x0589, 0x058a, 1},
+ {0x05be, 0x05c0, 2},
+ {0x05c3, 0x05c6, 3},
+ {0x05f3, 0x05f4, 1},
+ {0x0609, 0x060a, 1},
+ {0x060c, 0x060d, 1},
+ {0x061b, 0x061e, 3},
+ {0x061f, 0x066a, 75},
+ {0x066b, 0x066d, 1},
+ {0x06d4, 0x0700, 44},
+ {0x0701, 0x070d, 1},
+ {0x07f7, 0x07f9, 1},
+ {0x0830, 0x083e, 1},
+ {0x085e, 0x0964, 262},
+ {0x0965, 0x0970, 11},
+ {0x0af0, 0x0df4, 772},
+ {0x0e4f, 0x0e5a, 11},
+ {0x0e5b, 0x0f04, 169},
+ {0x0f05, 0x0f12, 1},
+ {0x0f14, 0x0f3a, 38},
+ {0x0f3b, 0x0f3d, 1},
+ {0x0f85, 0x0fd0, 75},
+ {0x0fd1, 0x0fd4, 1},
+ {0x0fd9, 0x0fda, 1},
+ {0x104a, 0x104f, 1},
+ {0x10fb, 0x1360, 613},
+ {0x1361, 0x1368, 1},
+ {0x1400, 0x166d, 621},
+ {0x166e, 0x169b, 45},
+ {0x169c, 0x16eb, 79},
+ {0x16ec, 0x16ed, 1},
+ {0x1735, 0x1736, 1},
+ {0x17d4, 0x17d6, 1},
+ {0x17d8, 0x17da, 1},
+ {0x1800, 0x180a, 1},
+ {0x1944, 0x1945, 1},
+ {0x1a1e, 0x1a1f, 1},
+ {0x1aa0, 0x1aa6, 1},
+ {0x1aa8, 0x1aad, 1},
+ {0x1b5a, 0x1b60, 1},
+ {0x1bfc, 0x1bff, 1},
+ {0x1c3b, 0x1c3f, 1},
+ {0x1c7e, 0x1c7f, 1},
+ {0x1cc0, 0x1cc7, 1},
+ {0x1cd3, 0x2010, 829},
+ {0x2011, 0x2027, 1},
+ {0x2030, 0x2043, 1},
+ {0x2045, 0x2051, 1},
+ {0x2053, 0x205e, 1},
+ {0x207d, 0x207e, 1},
+ {0x208d, 0x208e, 1},
+ {0x2308, 0x230b, 1},
+ {0x2329, 0x232a, 1},
+ {0x2768, 0x2775, 1},
+ {0x27c5, 0x27c6, 1},
+ {0x27e6, 0x27ef, 1},
+ {0x2983, 0x2998, 1},
+ {0x29d8, 0x29db, 1},
+ {0x29fc, 0x29fd, 1},
+ {0x2cf9, 0x2cfc, 1},
+ {0x2cfe, 0x2cff, 1},
+ {0x2d70, 0x2e00, 144},
+ {0x2e01, 0x2e2e, 1},
+ {0x2e30, 0x2e42, 1},
+ {0x3001, 0x3003, 1},
+ {0x3008, 0x3011, 1},
+ {0x3014, 0x301f, 1},
+ {0x3030, 0x303d, 13},
+ {0x30a0, 0x30fb, 91},
+ {0xa4fe, 0xa4ff, 1},
+ {0xa60d, 0xa60f, 1},
+ {0xa673, 0xa67e, 11},
+ {0xa6f2, 0xa6f7, 1},
+ {0xa874, 0xa877, 1},
+ {0xa8ce, 0xa8cf, 1},
+ {0xa8f8, 0xa8fa, 1},
+ {0xa92e, 0xa92f, 1},
+ {0xa95f, 0xa9c1, 98},
+ {0xa9c2, 0xa9cd, 1},
+ {0xa9de, 0xa9df, 1},
+ {0xaa5c, 0xaa5f, 1},
+ {0xaade, 0xaadf, 1},
+ {0xaaf0, 0xaaf1, 1},
+ {0xabeb, 0xfd3e, 20819},
+ {0xfd3f, 0xfe10, 209},
+ {0xfe11, 0xfe19, 1},
+ {0xfe30, 0xfe52, 1},
+ {0xfe54, 0xfe61, 1},
+ {0xfe63, 0xfe68, 5},
+ {0xfe6a, 0xfe6b, 1},
+ {0xff01, 0xff03, 1},
+ {0xff05, 0xff0a, 1},
+ {0xff0c, 0xff0f, 1},
+ {0xff1a, 0xff1b, 1},
+ {0xff1f, 0xff20, 1},
+ {0xff3b, 0xff3d, 1},
+ {0xff3f, 0xff5b, 28},
+ {0xff5d, 0xff5f, 2},
+ {0xff60, 0xff65, 1},
+ },
+ R32: []Range32{
+ {0x10100, 0x10102, 1},
+ {0x1039f, 0x103d0, 49},
+ {0x1056f, 0x10857, 744},
+ {0x1091f, 0x1093f, 32},
+ {0x10a50, 0x10a58, 1},
+ {0x10a7f, 0x10af0, 113},
+ {0x10af1, 0x10af6, 1},
+ {0x10b39, 0x10b3f, 1},
+ {0x10b99, 0x10b9c, 1},
+ {0x11047, 0x1104d, 1},
+ {0x110bb, 0x110bc, 1},
+ {0x110be, 0x110c1, 1},
+ {0x11140, 0x11143, 1},
+ {0x11174, 0x11175, 1},
+ {0x111c5, 0x111c8, 1},
+ {0x111cd, 0x11238, 107},
+ {0x11239, 0x1123d, 1},
+ {0x114c6, 0x115c1, 251},
+ {0x115c2, 0x115c9, 1},
+ {0x11641, 0x11643, 1},
+ {0x12470, 0x12474, 1},
+ {0x16a6e, 0x16a6f, 1},
+ {0x16af5, 0x16b37, 66},
+ {0x16b38, 0x16b3b, 1},
+ {0x16b44, 0x1bc9f, 20827},
+ },
+ LatinOffset: 11,
+}
+
+var _Pc = &RangeTable{
+ R16: []Range16{
+ {0x005f, 0x203f, 8160},
+ {0x2040, 0x2054, 20},
+ {0xfe33, 0xfe34, 1},
+ {0xfe4d, 0xfe4f, 1},
+ {0xff3f, 0xff3f, 1},
+ },
+}
+
+var _Pd = &RangeTable{
+ R16: []Range16{
+ {0x002d, 0x058a, 1373},
+ {0x05be, 0x1400, 3650},
+ {0x1806, 0x2010, 2058},
+ {0x2011, 0x2015, 1},
+ {0x2e17, 0x2e1a, 3},
+ {0x2e3a, 0x2e3b, 1},
+ {0x2e40, 0x301c, 476},
+ {0x3030, 0x30a0, 112},
+ {0xfe31, 0xfe32, 1},
+ {0xfe58, 0xfe63, 11},
+ {0xff0d, 0xff0d, 1},
+ },
+}
+
+var _Pe = &RangeTable{
+ R16: []Range16{
+ {0x0029, 0x005d, 52},
+ {0x007d, 0x0f3b, 3774},
+ {0x0f3d, 0x169c, 1887},
+ {0x2046, 0x207e, 56},
+ {0x208e, 0x2309, 635},
+ {0x230b, 0x232a, 31},
+ {0x2769, 0x2775, 2},
+ {0x27c6, 0x27e7, 33},
+ {0x27e9, 0x27ef, 2},
+ {0x2984, 0x2998, 2},
+ {0x29d9, 0x29db, 2},
+ {0x29fd, 0x2e23, 1062},
+ {0x2e25, 0x2e29, 2},
+ {0x3009, 0x3011, 2},
+ {0x3015, 0x301b, 2},
+ {0x301e, 0x301f, 1},
+ {0xfd3e, 0xfe18, 218},
+ {0xfe36, 0xfe44, 2},
+ {0xfe48, 0xfe5a, 18},
+ {0xfe5c, 0xfe5e, 2},
+ {0xff09, 0xff3d, 52},
+ {0xff5d, 0xff63, 3},
+ },
+ LatinOffset: 1,
+}
+
+var _Pf = &RangeTable{
+ R16: []Range16{
+ {0x00bb, 0x2019, 8030},
+ {0x201d, 0x203a, 29},
+ {0x2e03, 0x2e05, 2},
+ {0x2e0a, 0x2e0d, 3},
+ {0x2e1d, 0x2e21, 4},
+ },
+}
+
+var _Pi = &RangeTable{
+ R16: []Range16{
+ {0x00ab, 0x2018, 8045},
+ {0x201b, 0x201c, 1},
+ {0x201f, 0x2039, 26},
+ {0x2e02, 0x2e04, 2},
+ {0x2e09, 0x2e0c, 3},
+ {0x2e1c, 0x2e20, 4},
+ },
+}
+
+var _Po = &RangeTable{
+ R16: []Range16{
+ {0x0021, 0x0023, 1},
+ {0x0025, 0x0027, 1},
+ {0x002a, 0x002e, 2},
+ {0x002f, 0x003a, 11},
+ {0x003b, 0x003f, 4},
+ {0x0040, 0x005c, 28},
+ {0x00a1, 0x00a7, 6},
+ {0x00b6, 0x00b7, 1},
+ {0x00bf, 0x037e, 703},
+ {0x0387, 0x055a, 467},
+ {0x055b, 0x055f, 1},
+ {0x0589, 0x05c0, 55},
+ {0x05c3, 0x05c6, 3},
+ {0x05f3, 0x05f4, 1},
+ {0x0609, 0x060a, 1},
+ {0x060c, 0x060d, 1},
+ {0x061b, 0x061e, 3},
+ {0x061f, 0x066a, 75},
+ {0x066b, 0x066d, 1},
+ {0x06d4, 0x0700, 44},
+ {0x0701, 0x070d, 1},
+ {0x07f7, 0x07f9, 1},
+ {0x0830, 0x083e, 1},
+ {0x085e, 0x0964, 262},
+ {0x0965, 0x0970, 11},
+ {0x0af0, 0x0df4, 772},
+ {0x0e4f, 0x0e5a, 11},
+ {0x0e5b, 0x0f04, 169},
+ {0x0f05, 0x0f12, 1},
+ {0x0f14, 0x0f85, 113},
+ {0x0fd0, 0x0fd4, 1},
+ {0x0fd9, 0x0fda, 1},
+ {0x104a, 0x104f, 1},
+ {0x10fb, 0x1360, 613},
+ {0x1361, 0x1368, 1},
+ {0x166d, 0x166e, 1},
+ {0x16eb, 0x16ed, 1},
+ {0x1735, 0x1736, 1},
+ {0x17d4, 0x17d6, 1},
+ {0x17d8, 0x17da, 1},
+ {0x1800, 0x1805, 1},
+ {0x1807, 0x180a, 1},
+ {0x1944, 0x1945, 1},
+ {0x1a1e, 0x1a1f, 1},
+ {0x1aa0, 0x1aa6, 1},
+ {0x1aa8, 0x1aad, 1},
+ {0x1b5a, 0x1b60, 1},
+ {0x1bfc, 0x1bff, 1},
+ {0x1c3b, 0x1c3f, 1},
+ {0x1c7e, 0x1c7f, 1},
+ {0x1cc0, 0x1cc7, 1},
+ {0x1cd3, 0x2016, 835},
+ {0x2017, 0x2020, 9},
+ {0x2021, 0x2027, 1},
+ {0x2030, 0x2038, 1},
+ {0x203b, 0x203e, 1},
+ {0x2041, 0x2043, 1},
+ {0x2047, 0x2051, 1},
+ {0x2053, 0x2055, 2},
+ {0x2056, 0x205e, 1},
+ {0x2cf9, 0x2cfc, 1},
+ {0x2cfe, 0x2cff, 1},
+ {0x2d70, 0x2e00, 144},
+ {0x2e01, 0x2e06, 5},
+ {0x2e07, 0x2e08, 1},
+ {0x2e0b, 0x2e0e, 3},
+ {0x2e0f, 0x2e16, 1},
+ {0x2e18, 0x2e19, 1},
+ {0x2e1b, 0x2e1e, 3},
+ {0x2e1f, 0x2e2a, 11},
+ {0x2e2b, 0x2e2e, 1},
+ {0x2e30, 0x2e39, 1},
+ {0x2e3c, 0x2e3f, 1},
+ {0x2e41, 0x3001, 448},
+ {0x3002, 0x3003, 1},
+ {0x303d, 0x30fb, 190},
+ {0xa4fe, 0xa4ff, 1},
+ {0xa60d, 0xa60f, 1},
+ {0xa673, 0xa67e, 11},
+ {0xa6f2, 0xa6f7, 1},
+ {0xa874, 0xa877, 1},
+ {0xa8ce, 0xa8cf, 1},
+ {0xa8f8, 0xa8fa, 1},
+ {0xa92e, 0xa92f, 1},
+ {0xa95f, 0xa9c1, 98},
+ {0xa9c2, 0xa9cd, 1},
+ {0xa9de, 0xa9df, 1},
+ {0xaa5c, 0xaa5f, 1},
+ {0xaade, 0xaadf, 1},
+ {0xaaf0, 0xaaf1, 1},
+ {0xabeb, 0xfe10, 21029},
+ {0xfe11, 0xfe16, 1},
+ {0xfe19, 0xfe30, 23},
+ {0xfe45, 0xfe46, 1},
+ {0xfe49, 0xfe4c, 1},
+ {0xfe50, 0xfe52, 1},
+ {0xfe54, 0xfe57, 1},
+ {0xfe5f, 0xfe61, 1},
+ {0xfe68, 0xfe6a, 2},
+ {0xfe6b, 0xff01, 150},
+ {0xff02, 0xff03, 1},
+ {0xff05, 0xff07, 1},
+ {0xff0a, 0xff0e, 2},
+ {0xff0f, 0xff1a, 11},
+ {0xff1b, 0xff1f, 4},
+ {0xff20, 0xff3c, 28},
+ {0xff61, 0xff64, 3},
+ {0xff65, 0xff65, 1},
+ },
+ R32: []Range32{
+ {0x10100, 0x10100, 1},
+ {0x10101, 0x10102, 1},
+ {0x1039f, 0x103d0, 49},
+ {0x1056f, 0x10857, 744},
+ {0x1091f, 0x1093f, 32},
+ {0x10a50, 0x10a58, 1},
+ {0x10a7f, 0x10af0, 113},
+ {0x10af1, 0x10af6, 1},
+ {0x10b39, 0x10b3f, 1},
+ {0x10b99, 0x10b9c, 1},
+ {0x11047, 0x1104d, 1},
+ {0x110bb, 0x110bc, 1},
+ {0x110be, 0x110c1, 1},
+ {0x11140, 0x11143, 1},
+ {0x11174, 0x11175, 1},
+ {0x111c5, 0x111c8, 1},
+ {0x111cd, 0x11238, 107},
+ {0x11239, 0x1123d, 1},
+ {0x114c6, 0x115c1, 251},
+ {0x115c2, 0x115c9, 1},
+ {0x11641, 0x11643, 1},
+ {0x12470, 0x12474, 1},
+ {0x16a6e, 0x16a6f, 1},
+ {0x16af5, 0x16b37, 66},
+ {0x16b38, 0x16b3b, 1},
+ {0x16b44, 0x1bc9f, 20827},
+ },
+ LatinOffset: 8,
+}
+
+var _Ps = &RangeTable{
+ R16: []Range16{
+ {0x0028, 0x005b, 51},
+ {0x007b, 0x0f3a, 3775},
+ {0x0f3c, 0x169b, 1887},
+ {0x201a, 0x201e, 4},
+ {0x2045, 0x207d, 56},
+ {0x208d, 0x2308, 635},
+ {0x230a, 0x2329, 31},
+ {0x2768, 0x2774, 2},
+ {0x27c5, 0x27e6, 33},
+ {0x27e8, 0x27ee, 2},
+ {0x2983, 0x2997, 2},
+ {0x29d8, 0x29da, 2},
+ {0x29fc, 0x2e22, 1062},
+ {0x2e24, 0x2e28, 2},
+ {0x2e42, 0x3008, 454},
+ {0x300a, 0x3010, 2},
+ {0x3014, 0x301a, 2},
+ {0x301d, 0xfd3f, 52514},
+ {0xfe17, 0xfe35, 30},
+ {0xfe37, 0xfe43, 2},
+ {0xfe47, 0xfe59, 18},
+ {0xfe5b, 0xfe5d, 2},
+ {0xff08, 0xff3b, 51},
+ {0xff5b, 0xff5f, 4},
+ {0xff62, 0xff62, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _S = &RangeTable{
+ R16: []Range16{
+ {0x0024, 0x002b, 7},
+ {0x003c, 0x003e, 1},
+ {0x005e, 0x0060, 2},
+ {0x007c, 0x007e, 2},
+ {0x00a2, 0x00a6, 1},
+ {0x00a8, 0x00a9, 1},
+ {0x00ac, 0x00ae, 2},
+ {0x00af, 0x00b1, 1},
+ {0x00b4, 0x00b8, 4},
+ {0x00d7, 0x00f7, 32},
+ {0x02c2, 0x02c5, 1},
+ {0x02d2, 0x02df, 1},
+ {0x02e5, 0x02eb, 1},
+ {0x02ed, 0x02ef, 2},
+ {0x02f0, 0x02ff, 1},
+ {0x0375, 0x0384, 15},
+ {0x0385, 0x03f6, 113},
+ {0x0482, 0x058d, 267},
+ {0x058e, 0x058f, 1},
+ {0x0606, 0x0608, 1},
+ {0x060b, 0x060e, 3},
+ {0x060f, 0x06de, 207},
+ {0x06e9, 0x06fd, 20},
+ {0x06fe, 0x07f6, 248},
+ {0x09f2, 0x09f3, 1},
+ {0x09fa, 0x09fb, 1},
+ {0x0af1, 0x0b70, 127},
+ {0x0bf3, 0x0bfa, 1},
+ {0x0c7f, 0x0d79, 250},
+ {0x0e3f, 0x0f01, 194},
+ {0x0f02, 0x0f03, 1},
+ {0x0f13, 0x0f15, 2},
+ {0x0f16, 0x0f17, 1},
+ {0x0f1a, 0x0f1f, 1},
+ {0x0f34, 0x0f38, 2},
+ {0x0fbe, 0x0fc5, 1},
+ {0x0fc7, 0x0fcc, 1},
+ {0x0fce, 0x0fcf, 1},
+ {0x0fd5, 0x0fd8, 1},
+ {0x109e, 0x109f, 1},
+ {0x1390, 0x1399, 1},
+ {0x17db, 0x1940, 357},
+ {0x19de, 0x19ff, 1},
+ {0x1b61, 0x1b6a, 1},
+ {0x1b74, 0x1b7c, 1},
+ {0x1fbd, 0x1fbf, 2},
+ {0x1fc0, 0x1fc1, 1},
+ {0x1fcd, 0x1fcf, 1},
+ {0x1fdd, 0x1fdf, 1},
+ {0x1fed, 0x1fef, 1},
+ {0x1ffd, 0x1ffe, 1},
+ {0x2044, 0x2052, 14},
+ {0x207a, 0x207c, 1},
+ {0x208a, 0x208c, 1},
+ {0x20a0, 0x20bd, 1},
+ {0x2100, 0x2101, 1},
+ {0x2103, 0x2106, 1},
+ {0x2108, 0x2109, 1},
+ {0x2114, 0x2116, 2},
+ {0x2117, 0x2118, 1},
+ {0x211e, 0x2123, 1},
+ {0x2125, 0x2129, 2},
+ {0x212e, 0x213a, 12},
+ {0x213b, 0x2140, 5},
+ {0x2141, 0x2144, 1},
+ {0x214a, 0x214d, 1},
+ {0x214f, 0x2190, 65},
+ {0x2191, 0x2307, 1},
+ {0x230c, 0x2328, 1},
+ {0x232b, 0x23fa, 1},
+ {0x2400, 0x2426, 1},
+ {0x2440, 0x244a, 1},
+ {0x249c, 0x24e9, 1},
+ {0x2500, 0x2767, 1},
+ {0x2794, 0x27c4, 1},
+ {0x27c7, 0x27e5, 1},
+ {0x27f0, 0x2982, 1},
+ {0x2999, 0x29d7, 1},
+ {0x29dc, 0x29fb, 1},
+ {0x29fe, 0x2b73, 1},
+ {0x2b76, 0x2b95, 1},
+ {0x2b98, 0x2bb9, 1},
+ {0x2bbd, 0x2bc8, 1},
+ {0x2bca, 0x2bd1, 1},
+ {0x2ce5, 0x2cea, 1},
+ {0x2e80, 0x2e99, 1},
+ {0x2e9b, 0x2ef3, 1},
+ {0x2f00, 0x2fd5, 1},
+ {0x2ff0, 0x2ffb, 1},
+ {0x3004, 0x3012, 14},
+ {0x3013, 0x3020, 13},
+ {0x3036, 0x3037, 1},
+ {0x303e, 0x303f, 1},
+ {0x309b, 0x309c, 1},
+ {0x3190, 0x3191, 1},
+ {0x3196, 0x319f, 1},
+ {0x31c0, 0x31e3, 1},
+ {0x3200, 0x321e, 1},
+ {0x322a, 0x3247, 1},
+ {0x3250, 0x3260, 16},
+ {0x3261, 0x327f, 1},
+ {0x328a, 0x32b0, 1},
+ {0x32c0, 0x32fe, 1},
+ {0x3300, 0x33ff, 1},
+ {0x4dc0, 0x4dff, 1},
+ {0xa490, 0xa4c6, 1},
+ {0xa700, 0xa716, 1},
+ {0xa720, 0xa721, 1},
+ {0xa789, 0xa78a, 1},
+ {0xa828, 0xa82b, 1},
+ {0xa836, 0xa839, 1},
+ {0xaa77, 0xaa79, 1},
+ {0xab5b, 0xfb29, 20430},
+ {0xfbb2, 0xfbc1, 1},
+ {0xfdfc, 0xfdfd, 1},
+ {0xfe62, 0xfe64, 2},
+ {0xfe65, 0xfe66, 1},
+ {0xfe69, 0xff04, 155},
+ {0xff0b, 0xff1c, 17},
+ {0xff1d, 0xff1e, 1},
+ {0xff3e, 0xff40, 2},
+ {0xff5c, 0xff5e, 2},
+ {0xffe0, 0xffe6, 1},
+ {0xffe8, 0xffee, 1},
+ {0xfffc, 0xfffd, 1},
+ },
+ R32: []Range32{
+ {0x10137, 0x1013f, 1},
+ {0x10179, 0x10189, 1},
+ {0x1018c, 0x10190, 4},
+ {0x10191, 0x1019b, 1},
+ {0x101a0, 0x101d0, 48},
+ {0x101d1, 0x101fc, 1},
+ {0x10877, 0x10878, 1},
+ {0x10ac8, 0x16b3c, 24692},
+ {0x16b3d, 0x16b3f, 1},
+ {0x16b45, 0x1bc9c, 20823},
+ {0x1d000, 0x1d0f5, 1},
+ {0x1d100, 0x1d126, 1},
+ {0x1d129, 0x1d164, 1},
+ {0x1d16a, 0x1d16c, 1},
+ {0x1d183, 0x1d184, 1},
+ {0x1d18c, 0x1d1a9, 1},
+ {0x1d1ae, 0x1d1dd, 1},
+ {0x1d200, 0x1d241, 1},
+ {0x1d245, 0x1d300, 187},
+ {0x1d301, 0x1d356, 1},
+ {0x1d6c1, 0x1d6db, 26},
+ {0x1d6fb, 0x1d715, 26},
+ {0x1d735, 0x1d74f, 26},
+ {0x1d76f, 0x1d789, 26},
+ {0x1d7a9, 0x1d7c3, 26},
+ {0x1eef0, 0x1eef1, 1},
+ {0x1f000, 0x1f02b, 1},
+ {0x1f030, 0x1f093, 1},
+ {0x1f0a0, 0x1f0ae, 1},
+ {0x1f0b1, 0x1f0bf, 1},
+ {0x1f0c1, 0x1f0cf, 1},
+ {0x1f0d1, 0x1f0f5, 1},
+ {0x1f110, 0x1f12e, 1},
+ {0x1f130, 0x1f16b, 1},
+ {0x1f170, 0x1f19a, 1},
+ {0x1f1e6, 0x1f202, 1},
+ {0x1f210, 0x1f23a, 1},
+ {0x1f240, 0x1f248, 1},
+ {0x1f250, 0x1f251, 1},
+ {0x1f300, 0x1f32c, 1},
+ {0x1f330, 0x1f37d, 1},
+ {0x1f380, 0x1f3ce, 1},
+ {0x1f3d4, 0x1f3f7, 1},
+ {0x1f400, 0x1f4fe, 1},
+ {0x1f500, 0x1f54a, 1},
+ {0x1f550, 0x1f579, 1},
+ {0x1f57b, 0x1f5a3, 1},
+ {0x1f5a5, 0x1f642, 1},
+ {0x1f645, 0x1f6cf, 1},
+ {0x1f6e0, 0x1f6ec, 1},
+ {0x1f6f0, 0x1f6f3, 1},
+ {0x1f700, 0x1f773, 1},
+ {0x1f780, 0x1f7d4, 1},
+ {0x1f800, 0x1f80b, 1},
+ {0x1f810, 0x1f847, 1},
+ {0x1f850, 0x1f859, 1},
+ {0x1f860, 0x1f887, 1},
+ {0x1f890, 0x1f8ad, 1},
+ },
+ LatinOffset: 10,
+}
+
+var _Sc = &RangeTable{
+ R16: []Range16{
+ {0x0024, 0x00a2, 126},
+ {0x00a3, 0x00a5, 1},
+ {0x058f, 0x060b, 124},
+ {0x09f2, 0x09f3, 1},
+ {0x09fb, 0x0af1, 246},
+ {0x0bf9, 0x0e3f, 582},
+ {0x17db, 0x20a0, 2245},
+ {0x20a1, 0x20bd, 1},
+ {0xa838, 0xfdfc, 21956},
+ {0xfe69, 0xff04, 155},
+ {0xffe0, 0xffe1, 1},
+ {0xffe5, 0xffe6, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _Sk = &RangeTable{
+ R16: []Range16{
+ {0x005e, 0x0060, 2},
+ {0x00a8, 0x00af, 7},
+ {0x00b4, 0x00b8, 4},
+ {0x02c2, 0x02c5, 1},
+ {0x02d2, 0x02df, 1},
+ {0x02e5, 0x02eb, 1},
+ {0x02ed, 0x02ef, 2},
+ {0x02f0, 0x02ff, 1},
+ {0x0375, 0x0384, 15},
+ {0x0385, 0x1fbd, 7224},
+ {0x1fbf, 0x1fc1, 1},
+ {0x1fcd, 0x1fcf, 1},
+ {0x1fdd, 0x1fdf, 1},
+ {0x1fed, 0x1fef, 1},
+ {0x1ffd, 0x1ffe, 1},
+ {0x309b, 0x309c, 1},
+ {0xa700, 0xa716, 1},
+ {0xa720, 0xa721, 1},
+ {0xa789, 0xa78a, 1},
+ {0xab5b, 0xfbb2, 20567},
+ {0xfbb3, 0xfbc1, 1},
+ {0xff3e, 0xff40, 2},
+ {0xffe3, 0xffe3, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _Sm = &RangeTable{
+ R16: []Range16{
+ {0x002b, 0x003c, 17},
+ {0x003d, 0x003e, 1},
+ {0x007c, 0x007e, 2},
+ {0x00ac, 0x00b1, 5},
+ {0x00d7, 0x00f7, 32},
+ {0x03f6, 0x0606, 528},
+ {0x0607, 0x0608, 1},
+ {0x2044, 0x2052, 14},
+ {0x207a, 0x207c, 1},
+ {0x208a, 0x208c, 1},
+ {0x2118, 0x2140, 40},
+ {0x2141, 0x2144, 1},
+ {0x214b, 0x2190, 69},
+ {0x2191, 0x2194, 1},
+ {0x219a, 0x219b, 1},
+ {0x21a0, 0x21a6, 3},
+ {0x21ae, 0x21ce, 32},
+ {0x21cf, 0x21d2, 3},
+ {0x21d4, 0x21f4, 32},
+ {0x21f5, 0x22ff, 1},
+ {0x2320, 0x2321, 1},
+ {0x237c, 0x239b, 31},
+ {0x239c, 0x23b3, 1},
+ {0x23dc, 0x23e1, 1},
+ {0x25b7, 0x25c1, 10},
+ {0x25f8, 0x25ff, 1},
+ {0x266f, 0x27c0, 337},
+ {0x27c1, 0x27c4, 1},
+ {0x27c7, 0x27e5, 1},
+ {0x27f0, 0x27ff, 1},
+ {0x2900, 0x2982, 1},
+ {0x2999, 0x29d7, 1},
+ {0x29dc, 0x29fb, 1},
+ {0x29fe, 0x2aff, 1},
+ {0x2b30, 0x2b44, 1},
+ {0x2b47, 0x2b4c, 1},
+ {0xfb29, 0xfe62, 825},
+ {0xfe64, 0xfe66, 1},
+ {0xff0b, 0xff1c, 17},
+ {0xff1d, 0xff1e, 1},
+ {0xff5c, 0xff5e, 2},
+ {0xffe2, 0xffe9, 7},
+ {0xffea, 0xffec, 1},
+ },
+ R32: []Range32{
+ {0x1d6c1, 0x1d6db, 26},
+ {0x1d6fb, 0x1d715, 26},
+ {0x1d735, 0x1d74f, 26},
+ {0x1d76f, 0x1d789, 26},
+ {0x1d7a9, 0x1d7c3, 26},
+ {0x1eef0, 0x1eef1, 1},
+ },
+ LatinOffset: 5,
+}
+
+var _So = &RangeTable{
+ R16: []Range16{
+ {0x00a6, 0x00a9, 3},
+ {0x00ae, 0x00b0, 2},
+ {0x0482, 0x058d, 267},
+ {0x058e, 0x060e, 128},
+ {0x060f, 0x06de, 207},
+ {0x06e9, 0x06fd, 20},
+ {0x06fe, 0x07f6, 248},
+ {0x09fa, 0x0b70, 374},
+ {0x0bf3, 0x0bf8, 1},
+ {0x0bfa, 0x0c7f, 133},
+ {0x0d79, 0x0f01, 392},
+ {0x0f02, 0x0f03, 1},
+ {0x0f13, 0x0f15, 2},
+ {0x0f16, 0x0f17, 1},
+ {0x0f1a, 0x0f1f, 1},
+ {0x0f34, 0x0f38, 2},
+ {0x0fbe, 0x0fc5, 1},
+ {0x0fc7, 0x0fcc, 1},
+ {0x0fce, 0x0fcf, 1},
+ {0x0fd5, 0x0fd8, 1},
+ {0x109e, 0x109f, 1},
+ {0x1390, 0x1399, 1},
+ {0x1940, 0x19de, 158},
+ {0x19df, 0x19ff, 1},
+ {0x1b61, 0x1b6a, 1},
+ {0x1b74, 0x1b7c, 1},
+ {0x2100, 0x2101, 1},
+ {0x2103, 0x2106, 1},
+ {0x2108, 0x2109, 1},
+ {0x2114, 0x2116, 2},
+ {0x2117, 0x211e, 7},
+ {0x211f, 0x2123, 1},
+ {0x2125, 0x2129, 2},
+ {0x212e, 0x213a, 12},
+ {0x213b, 0x214a, 15},
+ {0x214c, 0x214d, 1},
+ {0x214f, 0x2195, 70},
+ {0x2196, 0x2199, 1},
+ {0x219c, 0x219f, 1},
+ {0x21a1, 0x21a2, 1},
+ {0x21a4, 0x21a5, 1},
+ {0x21a7, 0x21ad, 1},
+ {0x21af, 0x21cd, 1},
+ {0x21d0, 0x21d1, 1},
+ {0x21d3, 0x21d5, 2},
+ {0x21d6, 0x21f3, 1},
+ {0x2300, 0x2307, 1},
+ {0x230c, 0x231f, 1},
+ {0x2322, 0x2328, 1},
+ {0x232b, 0x237b, 1},
+ {0x237d, 0x239a, 1},
+ {0x23b4, 0x23db, 1},
+ {0x23e2, 0x23fa, 1},
+ {0x2400, 0x2426, 1},
+ {0x2440, 0x244a, 1},
+ {0x249c, 0x24e9, 1},
+ {0x2500, 0x25b6, 1},
+ {0x25b8, 0x25c0, 1},
+ {0x25c2, 0x25f7, 1},
+ {0x2600, 0x266e, 1},
+ {0x2670, 0x2767, 1},
+ {0x2794, 0x27bf, 1},
+ {0x2800, 0x28ff, 1},
+ {0x2b00, 0x2b2f, 1},
+ {0x2b45, 0x2b46, 1},
+ {0x2b4d, 0x2b73, 1},
+ {0x2b76, 0x2b95, 1},
+ {0x2b98, 0x2bb9, 1},
+ {0x2bbd, 0x2bc8, 1},
+ {0x2bca, 0x2bd1, 1},
+ {0x2ce5, 0x2cea, 1},
+ {0x2e80, 0x2e99, 1},
+ {0x2e9b, 0x2ef3, 1},
+ {0x2f00, 0x2fd5, 1},
+ {0x2ff0, 0x2ffb, 1},
+ {0x3004, 0x3012, 14},
+ {0x3013, 0x3020, 13},
+ {0x3036, 0x3037, 1},
+ {0x303e, 0x303f, 1},
+ {0x3190, 0x3191, 1},
+ {0x3196, 0x319f, 1},
+ {0x31c0, 0x31e3, 1},
+ {0x3200, 0x321e, 1},
+ {0x322a, 0x3247, 1},
+ {0x3250, 0x3260, 16},
+ {0x3261, 0x327f, 1},
+ {0x328a, 0x32b0, 1},
+ {0x32c0, 0x32fe, 1},
+ {0x3300, 0x33ff, 1},
+ {0x4dc0, 0x4dff, 1},
+ {0xa490, 0xa4c6, 1},
+ {0xa828, 0xa82b, 1},
+ {0xa836, 0xa837, 1},
+ {0xa839, 0xaa77, 574},
+ {0xaa78, 0xaa79, 1},
+ {0xfdfd, 0xffe4, 487},
+ {0xffe8, 0xffed, 5},
+ {0xffee, 0xfffc, 14},
+ {0xfffd, 0xfffd, 1},
+ },
+ R32: []Range32{
+ {0x10137, 0x10137, 1},
+ {0x10138, 0x1013f, 1},
+ {0x10179, 0x10189, 1},
+ {0x1018c, 0x10190, 4},
+ {0x10191, 0x1019b, 1},
+ {0x101a0, 0x101d0, 48},
+ {0x101d1, 0x101fc, 1},
+ {0x10877, 0x10878, 1},
+ {0x10ac8, 0x16b3c, 24692},
+ {0x16b3d, 0x16b3f, 1},
+ {0x16b45, 0x1bc9c, 20823},
+ {0x1d000, 0x1d0f5, 1},
+ {0x1d100, 0x1d126, 1},
+ {0x1d129, 0x1d164, 1},
+ {0x1d16a, 0x1d16c, 1},
+ {0x1d183, 0x1d184, 1},
+ {0x1d18c, 0x1d1a9, 1},
+ {0x1d1ae, 0x1d1dd, 1},
+ {0x1d200, 0x1d241, 1},
+ {0x1d245, 0x1d300, 187},
+ {0x1d301, 0x1d356, 1},
+ {0x1f000, 0x1f02b, 1},
+ {0x1f030, 0x1f093, 1},
+ {0x1f0a0, 0x1f0ae, 1},
+ {0x1f0b1, 0x1f0bf, 1},
+ {0x1f0c1, 0x1f0cf, 1},
+ {0x1f0d1, 0x1f0f5, 1},
+ {0x1f110, 0x1f12e, 1},
+ {0x1f130, 0x1f16b, 1},
+ {0x1f170, 0x1f19a, 1},
+ {0x1f1e6, 0x1f202, 1},
+ {0x1f210, 0x1f23a, 1},
+ {0x1f240, 0x1f248, 1},
+ {0x1f250, 0x1f251, 1},
+ {0x1f300, 0x1f32c, 1},
+ {0x1f330, 0x1f37d, 1},
+ {0x1f380, 0x1f3ce, 1},
+ {0x1f3d4, 0x1f3f7, 1},
+ {0x1f400, 0x1f4fe, 1},
+ {0x1f500, 0x1f54a, 1},
+ {0x1f550, 0x1f579, 1},
+ {0x1f57b, 0x1f5a3, 1},
+ {0x1f5a5, 0x1f642, 1},
+ {0x1f645, 0x1f6cf, 1},
+ {0x1f6e0, 0x1f6ec, 1},
+ {0x1f6f0, 0x1f6f3, 1},
+ {0x1f700, 0x1f773, 1},
+ {0x1f780, 0x1f7d4, 1},
+ {0x1f800, 0x1f80b, 1},
+ {0x1f810, 0x1f847, 1},
+ {0x1f850, 0x1f859, 1},
+ {0x1f860, 0x1f887, 1},
+ {0x1f890, 0x1f8ad, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _Z = &RangeTable{
+ R16: []Range16{
+ {0x0020, 0x00a0, 128},
+ {0x1680, 0x2000, 2432},
+ {0x2001, 0x200a, 1},
+ {0x2028, 0x2029, 1},
+ {0x202f, 0x205f, 48},
+ {0x3000, 0x3000, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Zl = &RangeTable{
+ R16: []Range16{
+ {0x2028, 0x2028, 1},
+ },
+}
+
+var _Zp = &RangeTable{
+ R16: []Range16{
+ {0x2029, 0x2029, 1},
+ },
+}
+
+var _Zs = &RangeTable{
+ R16: []Range16{
+ {0x0020, 0x00a0, 128},
+ {0x1680, 0x2000, 2432},
+ {0x2001, 0x200a, 1},
+ {0x202f, 0x205f, 48},
+ {0x3000, 0x3000, 1},
+ },
+ LatinOffset: 1,
+}
+
+// These variables have type *RangeTable.
+var (
+ Cc = _Cc // Cc is the set of Unicode characters in category Cc.
+ Cf = _Cf // Cf is the set of Unicode characters in category Cf.
+ Co = _Co // Co is the set of Unicode characters in category Co.
+ Cs = _Cs // Cs is the set of Unicode characters in category Cs.
+ Digit = _Nd // Digit is the set of Unicode characters with the "decimal digit" property.
+ Nd = _Nd // Nd is the set of Unicode characters in category Nd.
+ Letter = _L // Letter/L is the set of Unicode letters, category L.
+ L = _L
+ Lm = _Lm // Lm is the set of Unicode characters in category Lm.
+ Lo = _Lo // Lo is the set of Unicode characters in category Lo.
+ Lower = _Ll // Lower is the set of Unicode lower case letters.
+ Ll = _Ll // Ll is the set of Unicode characters in category Ll.
+ Mark = _M // Mark/M is the set of Unicode mark characters, category M.
+ M = _M
+ Mc = _Mc // Mc is the set of Unicode characters in category Mc.
+ Me = _Me // Me is the set of Unicode characters in category Me.
+ Mn = _Mn // Mn is the set of Unicode characters in category Mn.
+ Nl = _Nl // Nl is the set of Unicode characters in category Nl.
+ No = _No // No is the set of Unicode characters in category No.
+ Number = _N // Number/N is the set of Unicode number characters, category N.
+ N = _N
+ Other = _C // Other/C is the set of Unicode control and special characters, category C.
+ C = _C
+ Pc = _Pc // Pc is the set of Unicode characters in category Pc.
+ Pd = _Pd // Pd is the set of Unicode characters in category Pd.
+ Pe = _Pe // Pe is the set of Unicode characters in category Pe.
+ Pf = _Pf // Pf is the set of Unicode characters in category Pf.
+ Pi = _Pi // Pi is the set of Unicode characters in category Pi.
+ Po = _Po // Po is the set of Unicode characters in category Po.
+ Ps = _Ps // Ps is the set of Unicode characters in category Ps.
+ Punct = _P // Punct/P is the set of Unicode punctuation characters, category P.
+ P = _P
+ Sc = _Sc // Sc is the set of Unicode characters in category Sc.
+ Sk = _Sk // Sk is the set of Unicode characters in category Sk.
+ Sm = _Sm // Sm is the set of Unicode characters in category Sm.
+ So = _So // So is the set of Unicode characters in category So.
+ Space = _Z // Space/Z is the set of Unicode space characters, category Z.
+ Z = _Z
+ Symbol = _S // Symbol/S is the set of Unicode symbol characters, category S.
+ S = _S
+ Title = _Lt // Title is the set of Unicode title case letters.
+ Lt = _Lt // Lt is the set of Unicode characters in category Lt.
+ Upper = _Lu // Upper is the set of Unicode upper case letters.
+ Lu = _Lu // Lu is the set of Unicode characters in category Lu.
+ Zl = _Zl // Zl is the set of Unicode characters in category Zl.
+ Zp = _Zp // Zp is the set of Unicode characters in category Zp.
+ Zs = _Zs // Zs is the set of Unicode characters in category Zs.
+)
+
+// Generated by running
+// maketables --scripts=all --url=http://www.unicode.org/Public/7.0.0/ucd/
+// DO NOT EDIT
+
+// Scripts is the set of Unicode script tables.
+var Scripts = map[string]*RangeTable{
+ "Arabic": Arabic,
+ "Armenian": Armenian,
+ "Avestan": Avestan,
+ "Balinese": Balinese,
+ "Bamum": Bamum,
+ "Bassa_Vah": Bassa_Vah,
+ "Batak": Batak,
+ "Bengali": Bengali,
+ "Bopomofo": Bopomofo,
+ "Brahmi": Brahmi,
+ "Braille": Braille,
+ "Buginese": Buginese,
+ "Buhid": Buhid,
+ "Canadian_Aboriginal": Canadian_Aboriginal,
+ "Carian": Carian,
+ "Caucasian_Albanian": Caucasian_Albanian,
+ "Chakma": Chakma,
+ "Cham": Cham,
+ "Cherokee": Cherokee,
+ "Common": Common,
+ "Coptic": Coptic,
+ "Cuneiform": Cuneiform,
+ "Cypriot": Cypriot,
+ "Cyrillic": Cyrillic,
+ "Deseret": Deseret,
+ "Devanagari": Devanagari,
+ "Duployan": Duployan,
+ "Egyptian_Hieroglyphs": Egyptian_Hieroglyphs,
+ "Elbasan": Elbasan,
+ "Ethiopic": Ethiopic,
+ "Georgian": Georgian,
+ "Glagolitic": Glagolitic,
+ "Gothic": Gothic,
+ "Grantha": Grantha,
+ "Greek": Greek,
+ "Gujarati": Gujarati,
+ "Gurmukhi": Gurmukhi,
+ "Han": Han,
+ "Hangul": Hangul,
+ "Hanunoo": Hanunoo,
+ "Hebrew": Hebrew,
+ "Hiragana": Hiragana,
+ "Imperial_Aramaic": Imperial_Aramaic,
+ "Inherited": Inherited,
+ "Inscriptional_Pahlavi": Inscriptional_Pahlavi,
+ "Inscriptional_Parthian": Inscriptional_Parthian,
+ "Javanese": Javanese,
+ "Kaithi": Kaithi,
+ "Kannada": Kannada,
+ "Katakana": Katakana,
+ "Kayah_Li": Kayah_Li,
+ "Kharoshthi": Kharoshthi,
+ "Khmer": Khmer,
+ "Khojki": Khojki,
+ "Khudawadi": Khudawadi,
+ "Lao": Lao,
+ "Latin": Latin,
+ "Lepcha": Lepcha,
+ "Limbu": Limbu,
+ "Linear_A": Linear_A,
+ "Linear_B": Linear_B,
+ "Lisu": Lisu,
+ "Lycian": Lycian,
+ "Lydian": Lydian,
+ "Mahajani": Mahajani,
+ "Malayalam": Malayalam,
+ "Mandaic": Mandaic,
+ "Manichaean": Manichaean,
+ "Meetei_Mayek": Meetei_Mayek,
+ "Mende_Kikakui": Mende_Kikakui,
+ "Meroitic_Cursive": Meroitic_Cursive,
+ "Meroitic_Hieroglyphs": Meroitic_Hieroglyphs,
+ "Miao": Miao,
+ "Modi": Modi,
+ "Mongolian": Mongolian,
+ "Mro": Mro,
+ "Myanmar": Myanmar,
+ "Nabataean": Nabataean,
+ "New_Tai_Lue": New_Tai_Lue,
+ "Nko": Nko,
+ "Ogham": Ogham,
+ "Ol_Chiki": Ol_Chiki,
+ "Old_Italic": Old_Italic,
+ "Old_North_Arabian": Old_North_Arabian,
+ "Old_Permic": Old_Permic,
+ "Old_Persian": Old_Persian,
+ "Old_South_Arabian": Old_South_Arabian,
+ "Old_Turkic": Old_Turkic,
+ "Oriya": Oriya,
+ "Osmanya": Osmanya,
+ "Pahawh_Hmong": Pahawh_Hmong,
+ "Palmyrene": Palmyrene,
+ "Pau_Cin_Hau": Pau_Cin_Hau,
+ "Phags_Pa": Phags_Pa,
+ "Phoenician": Phoenician,
+ "Psalter_Pahlavi": Psalter_Pahlavi,
+ "Rejang": Rejang,
+ "Runic": Runic,
+ "Samaritan": Samaritan,
+ "Saurashtra": Saurashtra,
+ "Sharada": Sharada,
+ "Shavian": Shavian,
+ "Siddham": Siddham,
+ "Sinhala": Sinhala,
+ "Sora_Sompeng": Sora_Sompeng,
+ "Sundanese": Sundanese,
+ "Syloti_Nagri": Syloti_Nagri,
+ "Syriac": Syriac,
+ "Tagalog": Tagalog,
+ "Tagbanwa": Tagbanwa,
+ "Tai_Le": Tai_Le,
+ "Tai_Tham": Tai_Tham,
+ "Tai_Viet": Tai_Viet,
+ "Takri": Takri,
+ "Tamil": Tamil,
+ "Telugu": Telugu,
+ "Thaana": Thaana,
+ "Thai": Thai,
+ "Tibetan": Tibetan,
+ "Tifinagh": Tifinagh,
+ "Tirhuta": Tirhuta,
+ "Ugaritic": Ugaritic,
+ "Vai": Vai,
+ "Warang_Citi": Warang_Citi,
+ "Yi": Yi,
+}
+
+var _Arabic = &RangeTable{
+ R16: []Range16{
+ {0x0600, 0x0604, 1},
+ {0x0606, 0x060b, 1},
+ {0x060d, 0x061a, 1},
+ {0x061e, 0x061e, 1},
+ {0x0620, 0x063f, 1},
+ {0x0641, 0x064a, 1},
+ {0x0656, 0x065f, 1},
+ {0x066a, 0x066f, 1},
+ {0x0671, 0x06dc, 1},
+ {0x06de, 0x06ff, 1},
+ {0x0750, 0x077f, 1},
+ {0x08a0, 0x08b2, 1},
+ {0x08e4, 0x08ff, 1},
+ {0xfb50, 0xfbc1, 1},
+ {0xfbd3, 0xfd3d, 1},
+ {0xfd50, 0xfd8f, 1},
+ {0xfd92, 0xfdc7, 1},
+ {0xfdf0, 0xfdfd, 1},
+ {0xfe70, 0xfe74, 1},
+ {0xfe76, 0xfefc, 1},
+ },
+ R32: []Range32{
+ {0x10e60, 0x10e7e, 1},
+ {0x1ee00, 0x1ee03, 1},
+ {0x1ee05, 0x1ee1f, 1},
+ {0x1ee21, 0x1ee22, 1},
+ {0x1ee24, 0x1ee24, 1},
+ {0x1ee27, 0x1ee27, 1},
+ {0x1ee29, 0x1ee32, 1},
+ {0x1ee34, 0x1ee37, 1},
+ {0x1ee39, 0x1ee39, 1},
+ {0x1ee3b, 0x1ee3b, 1},
+ {0x1ee42, 0x1ee42, 1},
+ {0x1ee47, 0x1ee47, 1},
+ {0x1ee49, 0x1ee49, 1},
+ {0x1ee4b, 0x1ee4b, 1},
+ {0x1ee4d, 0x1ee4f, 1},
+ {0x1ee51, 0x1ee52, 1},
+ {0x1ee54, 0x1ee54, 1},
+ {0x1ee57, 0x1ee57, 1},
+ {0x1ee59, 0x1ee59, 1},
+ {0x1ee5b, 0x1ee5b, 1},
+ {0x1ee5d, 0x1ee5d, 1},
+ {0x1ee5f, 0x1ee5f, 1},
+ {0x1ee61, 0x1ee62, 1},
+ {0x1ee64, 0x1ee64, 1},
+ {0x1ee67, 0x1ee6a, 1},
+ {0x1ee6c, 0x1ee72, 1},
+ {0x1ee74, 0x1ee77, 1},
+ {0x1ee79, 0x1ee7c, 1},
+ {0x1ee7e, 0x1ee7e, 1},
+ {0x1ee80, 0x1ee89, 1},
+ {0x1ee8b, 0x1ee9b, 1},
+ {0x1eea1, 0x1eea3, 1},
+ {0x1eea5, 0x1eea9, 1},
+ {0x1eeab, 0x1eebb, 1},
+ {0x1eef0, 0x1eef1, 1},
+ },
+}
+
+var _Armenian = &RangeTable{
+ R16: []Range16{
+ {0x0531, 0x0556, 1},
+ {0x0559, 0x055f, 1},
+ {0x0561, 0x0587, 1},
+ {0x058a, 0x058a, 1},
+ {0x058d, 0x058f, 1},
+ {0xfb13, 0xfb17, 1},
+ },
+}
+
+var _Avestan = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10b00, 0x10b35, 1},
+ {0x10b39, 0x10b3f, 1},
+ },
+}
+
+var _Balinese = &RangeTable{
+ R16: []Range16{
+ {0x1b00, 0x1b4b, 1},
+ {0x1b50, 0x1b7c, 1},
+ },
+}
+
+var _Bamum = &RangeTable{
+ R16: []Range16{
+ {0xa6a0, 0xa6f7, 1},
+ },
+ R32: []Range32{
+ {0x16800, 0x16a38, 1},
+ },
+}
+
+var _Bassa_Vah = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x16ad0, 0x16aed, 1},
+ {0x16af0, 0x16af5, 1},
+ },
+}
+
+var _Batak = &RangeTable{
+ R16: []Range16{
+ {0x1bc0, 0x1bf3, 1},
+ {0x1bfc, 0x1bff, 1},
+ },
+}
+
+var _Bengali = &RangeTable{
+ R16: []Range16{
+ {0x0980, 0x0983, 1},
+ {0x0985, 0x098c, 1},
+ {0x098f, 0x0990, 1},
+ {0x0993, 0x09a8, 1},
+ {0x09aa, 0x09b0, 1},
+ {0x09b2, 0x09b2, 1},
+ {0x09b6, 0x09b9, 1},
+ {0x09bc, 0x09c4, 1},
+ {0x09c7, 0x09c8, 1},
+ {0x09cb, 0x09ce, 1},
+ {0x09d7, 0x09d7, 1},
+ {0x09dc, 0x09dd, 1},
+ {0x09df, 0x09e3, 1},
+ {0x09e6, 0x09fb, 1},
+ },
+}
+
+var _Bopomofo = &RangeTable{
+ R16: []Range16{
+ {0x02ea, 0x02eb, 1},
+ {0x3105, 0x312d, 1},
+ {0x31a0, 0x31ba, 1},
+ },
+}
+
+var _Brahmi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11000, 0x1104d, 1},
+ {0x11052, 0x1106f, 1},
+ {0x1107f, 0x1107f, 1},
+ },
+}
+
+var _Braille = &RangeTable{
+ R16: []Range16{
+ {0x2800, 0x28ff, 1},
+ },
+}
+
+var _Buginese = &RangeTable{
+ R16: []Range16{
+ {0x1a00, 0x1a1b, 1},
+ {0x1a1e, 0x1a1f, 1},
+ },
+}
+
+var _Buhid = &RangeTable{
+ R16: []Range16{
+ {0x1740, 0x1753, 1},
+ },
+}
+
+var _Canadian_Aboriginal = &RangeTable{
+ R16: []Range16{
+ {0x1400, 0x167f, 1},
+ {0x18b0, 0x18f5, 1},
+ },
+}
+
+var _Carian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x102a0, 0x102d0, 1},
+ },
+}
+
+var _Caucasian_Albanian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10530, 0x10563, 1},
+ {0x1056f, 0x1056f, 1},
+ },
+}
+
+var _Chakma = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11100, 0x11134, 1},
+ {0x11136, 0x11143, 1},
+ },
+}
+
+var _Cham = &RangeTable{
+ R16: []Range16{
+ {0xaa00, 0xaa36, 1},
+ {0xaa40, 0xaa4d, 1},
+ {0xaa50, 0xaa59, 1},
+ {0xaa5c, 0xaa5f, 1},
+ },
+}
+
+var _Cherokee = &RangeTable{
+ R16: []Range16{
+ {0x13a0, 0x13f4, 1},
+ },
+}
+
+var _Common = &RangeTable{
+ R16: []Range16{
+ {0x0000, 0x0040, 1},
+ {0x005b, 0x0060, 1},
+ {0x007b, 0x00a9, 1},
+ {0x00ab, 0x00b9, 1},
+ {0x00bb, 0x00bf, 1},
+ {0x00d7, 0x00d7, 1},
+ {0x00f7, 0x00f7, 1},
+ {0x02b9, 0x02df, 1},
+ {0x02e5, 0x02e9, 1},
+ {0x02ec, 0x02ff, 1},
+ {0x0374, 0x0374, 1},
+ {0x037e, 0x037e, 1},
+ {0x0385, 0x0385, 1},
+ {0x0387, 0x0387, 1},
+ {0x0589, 0x0589, 1},
+ {0x0605, 0x0605, 1},
+ {0x060c, 0x060c, 1},
+ {0x061b, 0x061c, 1},
+ {0x061f, 0x061f, 1},
+ {0x0640, 0x0640, 1},
+ {0x0660, 0x0669, 1},
+ {0x06dd, 0x06dd, 1},
+ {0x0964, 0x0965, 1},
+ {0x0e3f, 0x0e3f, 1},
+ {0x0fd5, 0x0fd8, 1},
+ {0x10fb, 0x10fb, 1},
+ {0x16eb, 0x16ed, 1},
+ {0x1735, 0x1736, 1},
+ {0x1802, 0x1803, 1},
+ {0x1805, 0x1805, 1},
+ {0x1cd3, 0x1cd3, 1},
+ {0x1ce1, 0x1ce1, 1},
+ {0x1ce9, 0x1cec, 1},
+ {0x1cee, 0x1cf3, 1},
+ {0x1cf5, 0x1cf6, 1},
+ {0x2000, 0x200b, 1},
+ {0x200e, 0x2064, 1},
+ {0x2066, 0x2070, 1},
+ {0x2074, 0x207e, 1},
+ {0x2080, 0x208e, 1},
+ {0x20a0, 0x20bd, 1},
+ {0x2100, 0x2125, 1},
+ {0x2127, 0x2129, 1},
+ {0x212c, 0x2131, 1},
+ {0x2133, 0x214d, 1},
+ {0x214f, 0x215f, 1},
+ {0x2189, 0x2189, 1},
+ {0x2190, 0x23fa, 1},
+ {0x2400, 0x2426, 1},
+ {0x2440, 0x244a, 1},
+ {0x2460, 0x27ff, 1},
+ {0x2900, 0x2b73, 1},
+ {0x2b76, 0x2b95, 1},
+ {0x2b98, 0x2bb9, 1},
+ {0x2bbd, 0x2bc8, 1},
+ {0x2bca, 0x2bd1, 1},
+ {0x2e00, 0x2e42, 1},
+ {0x2ff0, 0x2ffb, 1},
+ {0x3000, 0x3004, 1},
+ {0x3006, 0x3006, 1},
+ {0x3008, 0x3020, 1},
+ {0x3030, 0x3037, 1},
+ {0x303c, 0x303f, 1},
+ {0x309b, 0x309c, 1},
+ {0x30a0, 0x30a0, 1},
+ {0x30fb, 0x30fc, 1},
+ {0x3190, 0x319f, 1},
+ {0x31c0, 0x31e3, 1},
+ {0x3220, 0x325f, 1},
+ {0x327f, 0x32cf, 1},
+ {0x3358, 0x33ff, 1},
+ {0x4dc0, 0x4dff, 1},
+ {0xa700, 0xa721, 1},
+ {0xa788, 0xa78a, 1},
+ {0xa830, 0xa839, 1},
+ {0xa92e, 0xa92e, 1},
+ {0xa9cf, 0xa9cf, 1},
+ {0xab5b, 0xab5b, 1},
+ {0xfd3e, 0xfd3f, 1},
+ {0xfe10, 0xfe19, 1},
+ {0xfe30, 0xfe52, 1},
+ {0xfe54, 0xfe66, 1},
+ {0xfe68, 0xfe6b, 1},
+ {0xfeff, 0xfeff, 1},
+ {0xff01, 0xff20, 1},
+ {0xff3b, 0xff40, 1},
+ {0xff5b, 0xff65, 1},
+ {0xff70, 0xff70, 1},
+ {0xff9e, 0xff9f, 1},
+ {0xffe0, 0xffe6, 1},
+ {0xffe8, 0xffee, 1},
+ {0xfff9, 0xfffd, 1},
+ },
+ R32: []Range32{
+ {0x10100, 0x10102, 1},
+ {0x10107, 0x10133, 1},
+ {0x10137, 0x1013f, 1},
+ {0x10190, 0x1019b, 1},
+ {0x101d0, 0x101fc, 1},
+ {0x102e1, 0x102fb, 1},
+ {0x1bca0, 0x1bca3, 1},
+ {0x1d000, 0x1d0f5, 1},
+ {0x1d100, 0x1d126, 1},
+ {0x1d129, 0x1d166, 1},
+ {0x1d16a, 0x1d17a, 1},
+ {0x1d183, 0x1d184, 1},
+ {0x1d18c, 0x1d1a9, 1},
+ {0x1d1ae, 0x1d1dd, 1},
+ {0x1d300, 0x1d356, 1},
+ {0x1d360, 0x1d371, 1},
+ {0x1d400, 0x1d454, 1},
+ {0x1d456, 0x1d49c, 1},
+ {0x1d49e, 0x1d49f, 1},
+ {0x1d4a2, 0x1d4a2, 1},
+ {0x1d4a5, 0x1d4a6, 1},
+ {0x1d4a9, 0x1d4ac, 1},
+ {0x1d4ae, 0x1d4b9, 1},
+ {0x1d4bb, 0x1d4bb, 1},
+ {0x1d4bd, 0x1d4c3, 1},
+ {0x1d4c5, 0x1d505, 1},
+ {0x1d507, 0x1d50a, 1},
+ {0x1d50d, 0x1d514, 1},
+ {0x1d516, 0x1d51c, 1},
+ {0x1d51e, 0x1d539, 1},
+ {0x1d53b, 0x1d53e, 1},
+ {0x1d540, 0x1d544, 1},
+ {0x1d546, 0x1d546, 1},
+ {0x1d54a, 0x1d550, 1},
+ {0x1d552, 0x1d6a5, 1},
+ {0x1d6a8, 0x1d7cb, 1},
+ {0x1d7ce, 0x1d7ff, 1},
+ {0x1f000, 0x1f02b, 1},
+ {0x1f030, 0x1f093, 1},
+ {0x1f0a0, 0x1f0ae, 1},
+ {0x1f0b1, 0x1f0bf, 1},
+ {0x1f0c1, 0x1f0cf, 1},
+ {0x1f0d1, 0x1f0f5, 1},
+ {0x1f100, 0x1f10c, 1},
+ {0x1f110, 0x1f12e, 1},
+ {0x1f130, 0x1f16b, 1},
+ {0x1f170, 0x1f19a, 1},
+ {0x1f1e6, 0x1f1ff, 1},
+ {0x1f201, 0x1f202, 1},
+ {0x1f210, 0x1f23a, 1},
+ {0x1f240, 0x1f248, 1},
+ {0x1f250, 0x1f251, 1},
+ {0x1f300, 0x1f32c, 1},
+ {0x1f330, 0x1f37d, 1},
+ {0x1f380, 0x1f3ce, 1},
+ {0x1f3d4, 0x1f3f7, 1},
+ {0x1f400, 0x1f4fe, 1},
+ {0x1f500, 0x1f54a, 1},
+ {0x1f550, 0x1f579, 1},
+ {0x1f57b, 0x1f5a3, 1},
+ {0x1f5a5, 0x1f642, 1},
+ {0x1f645, 0x1f6cf, 1},
+ {0x1f6e0, 0x1f6ec, 1},
+ {0x1f6f0, 0x1f6f3, 1},
+ {0x1f700, 0x1f773, 1},
+ {0x1f780, 0x1f7d4, 1},
+ {0x1f800, 0x1f80b, 1},
+ {0x1f810, 0x1f847, 1},
+ {0x1f850, 0x1f859, 1},
+ {0x1f860, 0x1f887, 1},
+ {0x1f890, 0x1f8ad, 1},
+ {0xe0001, 0xe0001, 1},
+ {0xe0020, 0xe007f, 1},
+ },
+ LatinOffset: 7,
+}
+
+var _Coptic = &RangeTable{
+ R16: []Range16{
+ {0x03e2, 0x03ef, 1},
+ {0x2c80, 0x2cf3, 1},
+ {0x2cf9, 0x2cff, 1},
+ },
+}
+
+var _Cuneiform = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x12000, 0x12398, 1},
+ {0x12400, 0x1246e, 1},
+ {0x12470, 0x12474, 1},
+ },
+}
+
+var _Cypriot = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10800, 0x10805, 1},
+ {0x10808, 0x10808, 1},
+ {0x1080a, 0x10835, 1},
+ {0x10837, 0x10838, 1},
+ {0x1083c, 0x1083c, 1},
+ {0x1083f, 0x1083f, 1},
+ },
+}
+
+var _Cyrillic = &RangeTable{
+ R16: []Range16{
+ {0x0400, 0x0484, 1},
+ {0x0487, 0x052f, 1},
+ {0x1d2b, 0x1d2b, 1},
+ {0x1d78, 0x1d78, 1},
+ {0x2de0, 0x2dff, 1},
+ {0xa640, 0xa69d, 1},
+ {0xa69f, 0xa69f, 1},
+ },
+}
+
+var _Deseret = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10400, 0x1044f, 1},
+ },
+}
+
+var _Devanagari = &RangeTable{
+ R16: []Range16{
+ {0x0900, 0x0950, 1},
+ {0x0953, 0x0963, 1},
+ {0x0966, 0x097f, 1},
+ {0xa8e0, 0xa8fb, 1},
+ },
+}
+
+var _Duployan = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x1bc00, 0x1bc6a, 1},
+ {0x1bc70, 0x1bc7c, 1},
+ {0x1bc80, 0x1bc88, 1},
+ {0x1bc90, 0x1bc99, 1},
+ {0x1bc9c, 0x1bc9f, 1},
+ },
+}
+
+var _Egyptian_Hieroglyphs = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x13000, 0x1342e, 1},
+ },
+}
+
+var _Elbasan = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10500, 0x10527, 1},
+ },
+}
+
+var _Ethiopic = &RangeTable{
+ R16: []Range16{
+ {0x1200, 0x1248, 1},
+ {0x124a, 0x124d, 1},
+ {0x1250, 0x1256, 1},
+ {0x1258, 0x1258, 1},
+ {0x125a, 0x125d, 1},
+ {0x1260, 0x1288, 1},
+ {0x128a, 0x128d, 1},
+ {0x1290, 0x12b0, 1},
+ {0x12b2, 0x12b5, 1},
+ {0x12b8, 0x12be, 1},
+ {0x12c0, 0x12c0, 1},
+ {0x12c2, 0x12c5, 1},
+ {0x12c8, 0x12d6, 1},
+ {0x12d8, 0x1310, 1},
+ {0x1312, 0x1315, 1},
+ {0x1318, 0x135a, 1},
+ {0x135d, 0x137c, 1},
+ {0x1380, 0x1399, 1},
+ {0x2d80, 0x2d96, 1},
+ {0x2da0, 0x2da6, 1},
+ {0x2da8, 0x2dae, 1},
+ {0x2db0, 0x2db6, 1},
+ {0x2db8, 0x2dbe, 1},
+ {0x2dc0, 0x2dc6, 1},
+ {0x2dc8, 0x2dce, 1},
+ {0x2dd0, 0x2dd6, 1},
+ {0x2dd8, 0x2dde, 1},
+ {0xab01, 0xab06, 1},
+ {0xab09, 0xab0e, 1},
+ {0xab11, 0xab16, 1},
+ {0xab20, 0xab26, 1},
+ {0xab28, 0xab2e, 1},
+ },
+}
+
+var _Georgian = &RangeTable{
+ R16: []Range16{
+ {0x10a0, 0x10c5, 1},
+ {0x10c7, 0x10c7, 1},
+ {0x10cd, 0x10cd, 1},
+ {0x10d0, 0x10fa, 1},
+ {0x10fc, 0x10ff, 1},
+ {0x2d00, 0x2d25, 1},
+ {0x2d27, 0x2d27, 1},
+ {0x2d2d, 0x2d2d, 1},
+ },
+}
+
+var _Glagolitic = &RangeTable{
+ R16: []Range16{
+ {0x2c00, 0x2c2e, 1},
+ {0x2c30, 0x2c5e, 1},
+ },
+}
+
+var _Gothic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10330, 0x1034a, 1},
+ },
+}
+
+var _Grantha = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11301, 0x11303, 1},
+ {0x11305, 0x1130c, 1},
+ {0x1130f, 0x11310, 1},
+ {0x11313, 0x11328, 1},
+ {0x1132a, 0x11330, 1},
+ {0x11332, 0x11333, 1},
+ {0x11335, 0x11339, 1},
+ {0x1133c, 0x11344, 1},
+ {0x11347, 0x11348, 1},
+ {0x1134b, 0x1134d, 1},
+ {0x11357, 0x11357, 1},
+ {0x1135d, 0x11363, 1},
+ {0x11366, 0x1136c, 1},
+ {0x11370, 0x11374, 1},
+ },
+}
+
+var _Greek = &RangeTable{
+ R16: []Range16{
+ {0x0370, 0x0373, 1},
+ {0x0375, 0x0377, 1},
+ {0x037a, 0x037d, 1},
+ {0x037f, 0x037f, 1},
+ {0x0384, 0x0384, 1},
+ {0x0386, 0x0386, 1},
+ {0x0388, 0x038a, 1},
+ {0x038c, 0x038c, 1},
+ {0x038e, 0x03a1, 1},
+ {0x03a3, 0x03e1, 1},
+ {0x03f0, 0x03ff, 1},
+ {0x1d26, 0x1d2a, 1},
+ {0x1d5d, 0x1d61, 1},
+ {0x1d66, 0x1d6a, 1},
+ {0x1dbf, 0x1dbf, 1},
+ {0x1f00, 0x1f15, 1},
+ {0x1f18, 0x1f1d, 1},
+ {0x1f20, 0x1f45, 1},
+ {0x1f48, 0x1f4d, 1},
+ {0x1f50, 0x1f57, 1},
+ {0x1f59, 0x1f59, 1},
+ {0x1f5b, 0x1f5b, 1},
+ {0x1f5d, 0x1f5d, 1},
+ {0x1f5f, 0x1f7d, 1},
+ {0x1f80, 0x1fb4, 1},
+ {0x1fb6, 0x1fc4, 1},
+ {0x1fc6, 0x1fd3, 1},
+ {0x1fd6, 0x1fdb, 1},
+ {0x1fdd, 0x1fef, 1},
+ {0x1ff2, 0x1ff4, 1},
+ {0x1ff6, 0x1ffe, 1},
+ {0x2126, 0x2126, 1},
+ {0xab65, 0xab65, 1},
+ },
+ R32: []Range32{
+ {0x10140, 0x1018c, 1},
+ {0x101a0, 0x101a0, 1},
+ {0x1d200, 0x1d245, 1},
+ },
+}
+
+var _Gujarati = &RangeTable{
+ R16: []Range16{
+ {0x0a81, 0x0a83, 1},
+ {0x0a85, 0x0a8d, 1},
+ {0x0a8f, 0x0a91, 1},
+ {0x0a93, 0x0aa8, 1},
+ {0x0aaa, 0x0ab0, 1},
+ {0x0ab2, 0x0ab3, 1},
+ {0x0ab5, 0x0ab9, 1},
+ {0x0abc, 0x0ac5, 1},
+ {0x0ac7, 0x0ac9, 1},
+ {0x0acb, 0x0acd, 1},
+ {0x0ad0, 0x0ad0, 1},
+ {0x0ae0, 0x0ae3, 1},
+ {0x0ae6, 0x0af1, 1},
+ },
+}
+
+var _Gurmukhi = &RangeTable{
+ R16: []Range16{
+ {0x0a01, 0x0a03, 1},
+ {0x0a05, 0x0a0a, 1},
+ {0x0a0f, 0x0a10, 1},
+ {0x0a13, 0x0a28, 1},
+ {0x0a2a, 0x0a30, 1},
+ {0x0a32, 0x0a33, 1},
+ {0x0a35, 0x0a36, 1},
+ {0x0a38, 0x0a39, 1},
+ {0x0a3c, 0x0a3c, 1},
+ {0x0a3e, 0x0a42, 1},
+ {0x0a47, 0x0a48, 1},
+ {0x0a4b, 0x0a4d, 1},
+ {0x0a51, 0x0a51, 1},
+ {0x0a59, 0x0a5c, 1},
+ {0x0a5e, 0x0a5e, 1},
+ {0x0a66, 0x0a75, 1},
+ },
+}
+
+var _Han = &RangeTable{
+ R16: []Range16{
+ {0x2e80, 0x2e99, 1},
+ {0x2e9b, 0x2ef3, 1},
+ {0x2f00, 0x2fd5, 1},
+ {0x3005, 0x3005, 1},
+ {0x3007, 0x3007, 1},
+ {0x3021, 0x3029, 1},
+ {0x3038, 0x303b, 1},
+ {0x3400, 0x4db5, 1},
+ {0x4e00, 0x9fcc, 1},
+ {0xf900, 0xfa6d, 1},
+ {0xfa70, 0xfad9, 1},
+ },
+ R32: []Range32{
+ {0x20000, 0x2a6d6, 1},
+ {0x2a700, 0x2b734, 1},
+ {0x2b740, 0x2b81d, 1},
+ {0x2f800, 0x2fa1d, 1},
+ },
+}
+
+var _Hangul = &RangeTable{
+ R16: []Range16{
+ {0x1100, 0x11ff, 1},
+ {0x302e, 0x302f, 1},
+ {0x3131, 0x318e, 1},
+ {0x3200, 0x321e, 1},
+ {0x3260, 0x327e, 1},
+ {0xa960, 0xa97c, 1},
+ {0xac00, 0xd7a3, 1},
+ {0xd7b0, 0xd7c6, 1},
+ {0xd7cb, 0xd7fb, 1},
+ {0xffa0, 0xffbe, 1},
+ {0xffc2, 0xffc7, 1},
+ {0xffca, 0xffcf, 1},
+ {0xffd2, 0xffd7, 1},
+ {0xffda, 0xffdc, 1},
+ },
+}
+
+var _Hanunoo = &RangeTable{
+ R16: []Range16{
+ {0x1720, 0x1734, 1},
+ },
+}
+
+var _Hebrew = &RangeTable{
+ R16: []Range16{
+ {0x0591, 0x05c7, 1},
+ {0x05d0, 0x05ea, 1},
+ {0x05f0, 0x05f4, 1},
+ {0xfb1d, 0xfb36, 1},
+ {0xfb38, 0xfb3c, 1},
+ {0xfb3e, 0xfb3e, 1},
+ {0xfb40, 0xfb41, 1},
+ {0xfb43, 0xfb44, 1},
+ {0xfb46, 0xfb4f, 1},
+ },
+}
+
+var _Hiragana = &RangeTable{
+ R16: []Range16{
+ {0x3041, 0x3096, 1},
+ {0x309d, 0x309f, 1},
+ },
+ R32: []Range32{
+ {0x1b001, 0x1b001, 1},
+ {0x1f200, 0x1f200, 1},
+ },
+}
+
+var _Imperial_Aramaic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10840, 0x10855, 1},
+ {0x10857, 0x1085f, 1},
+ },
+}
+
+var _Inherited = &RangeTable{
+ R16: []Range16{
+ {0x0300, 0x036f, 1},
+ {0x0485, 0x0486, 1},
+ {0x064b, 0x0655, 1},
+ {0x0670, 0x0670, 1},
+ {0x0951, 0x0952, 1},
+ {0x1ab0, 0x1abe, 1},
+ {0x1cd0, 0x1cd2, 1},
+ {0x1cd4, 0x1ce0, 1},
+ {0x1ce2, 0x1ce8, 1},
+ {0x1ced, 0x1ced, 1},
+ {0x1cf4, 0x1cf4, 1},
+ {0x1cf8, 0x1cf9, 1},
+ {0x1dc0, 0x1df5, 1},
+ {0x1dfc, 0x1dff, 1},
+ {0x200c, 0x200d, 1},
+ {0x20d0, 0x20f0, 1},
+ {0x302a, 0x302d, 1},
+ {0x3099, 0x309a, 1},
+ {0xfe00, 0xfe0f, 1},
+ {0xfe20, 0xfe2d, 1},
+ },
+ R32: []Range32{
+ {0x101fd, 0x101fd, 1},
+ {0x102e0, 0x102e0, 1},
+ {0x1d167, 0x1d169, 1},
+ {0x1d17b, 0x1d182, 1},
+ {0x1d185, 0x1d18b, 1},
+ {0x1d1aa, 0x1d1ad, 1},
+ {0xe0100, 0xe01ef, 1},
+ },
+}
+
+var _Inscriptional_Pahlavi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10b60, 0x10b72, 1},
+ {0x10b78, 0x10b7f, 1},
+ },
+}
+
+var _Inscriptional_Parthian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10b40, 0x10b55, 1},
+ {0x10b58, 0x10b5f, 1},
+ },
+}
+
+var _Javanese = &RangeTable{
+ R16: []Range16{
+ {0xa980, 0xa9cd, 1},
+ {0xa9d0, 0xa9d9, 1},
+ {0xa9de, 0xa9df, 1},
+ },
+}
+
+var _Kaithi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11080, 0x110c1, 1},
+ },
+}
+
+var _Kannada = &RangeTable{
+ R16: []Range16{
+ {0x0c81, 0x0c83, 1},
+ {0x0c85, 0x0c8c, 1},
+ {0x0c8e, 0x0c90, 1},
+ {0x0c92, 0x0ca8, 1},
+ {0x0caa, 0x0cb3, 1},
+ {0x0cb5, 0x0cb9, 1},
+ {0x0cbc, 0x0cc4, 1},
+ {0x0cc6, 0x0cc8, 1},
+ {0x0cca, 0x0ccd, 1},
+ {0x0cd5, 0x0cd6, 1},
+ {0x0cde, 0x0cde, 1},
+ {0x0ce0, 0x0ce3, 1},
+ {0x0ce6, 0x0cef, 1},
+ {0x0cf1, 0x0cf2, 1},
+ },
+}
+
+var _Katakana = &RangeTable{
+ R16: []Range16{
+ {0x30a1, 0x30fa, 1},
+ {0x30fd, 0x30ff, 1},
+ {0x31f0, 0x31ff, 1},
+ {0x32d0, 0x32fe, 1},
+ {0x3300, 0x3357, 1},
+ {0xff66, 0xff6f, 1},
+ {0xff71, 0xff9d, 1},
+ },
+ R32: []Range32{
+ {0x1b000, 0x1b000, 1},
+ },
+}
+
+var _Kayah_Li = &RangeTable{
+ R16: []Range16{
+ {0xa900, 0xa92d, 1},
+ {0xa92f, 0xa92f, 1},
+ },
+}
+
+var _Kharoshthi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10a00, 0x10a03, 1},
+ {0x10a05, 0x10a06, 1},
+ {0x10a0c, 0x10a13, 1},
+ {0x10a15, 0x10a17, 1},
+ {0x10a19, 0x10a33, 1},
+ {0x10a38, 0x10a3a, 1},
+ {0x10a3f, 0x10a47, 1},
+ {0x10a50, 0x10a58, 1},
+ },
+}
+
+var _Khmer = &RangeTable{
+ R16: []Range16{
+ {0x1780, 0x17dd, 1},
+ {0x17e0, 0x17e9, 1},
+ {0x17f0, 0x17f9, 1},
+ {0x19e0, 0x19ff, 1},
+ },
+}
+
+var _Khojki = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11200, 0x11211, 1},
+ {0x11213, 0x1123d, 1},
+ },
+}
+
+var _Khudawadi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x112b0, 0x112ea, 1},
+ {0x112f0, 0x112f9, 1},
+ },
+}
+
+var _Lao = &RangeTable{
+ R16: []Range16{
+ {0x0e81, 0x0e82, 1},
+ {0x0e84, 0x0e84, 1},
+ {0x0e87, 0x0e88, 1},
+ {0x0e8a, 0x0e8a, 1},
+ {0x0e8d, 0x0e8d, 1},
+ {0x0e94, 0x0e97, 1},
+ {0x0e99, 0x0e9f, 1},
+ {0x0ea1, 0x0ea3, 1},
+ {0x0ea5, 0x0ea5, 1},
+ {0x0ea7, 0x0ea7, 1},
+ {0x0eaa, 0x0eab, 1},
+ {0x0ead, 0x0eb9, 1},
+ {0x0ebb, 0x0ebd, 1},
+ {0x0ec0, 0x0ec4, 1},
+ {0x0ec6, 0x0ec6, 1},
+ {0x0ec8, 0x0ecd, 1},
+ {0x0ed0, 0x0ed9, 1},
+ {0x0edc, 0x0edf, 1},
+ },
+}
+
+var _Latin = &RangeTable{
+ R16: []Range16{
+ {0x0041, 0x005a, 1},
+ {0x0061, 0x007a, 1},
+ {0x00aa, 0x00aa, 1},
+ {0x00ba, 0x00ba, 1},
+ {0x00c0, 0x00d6, 1},
+ {0x00d8, 0x00f6, 1},
+ {0x00f8, 0x02b8, 1},
+ {0x02e0, 0x02e4, 1},
+ {0x1d00, 0x1d25, 1},
+ {0x1d2c, 0x1d5c, 1},
+ {0x1d62, 0x1d65, 1},
+ {0x1d6b, 0x1d77, 1},
+ {0x1d79, 0x1dbe, 1},
+ {0x1e00, 0x1eff, 1},
+ {0x2071, 0x2071, 1},
+ {0x207f, 0x207f, 1},
+ {0x2090, 0x209c, 1},
+ {0x212a, 0x212b, 1},
+ {0x2132, 0x2132, 1},
+ {0x214e, 0x214e, 1},
+ {0x2160, 0x2188, 1},
+ {0x2c60, 0x2c7f, 1},
+ {0xa722, 0xa787, 1},
+ {0xa78b, 0xa78e, 1},
+ {0xa790, 0xa7ad, 1},
+ {0xa7b0, 0xa7b1, 1},
+ {0xa7f7, 0xa7ff, 1},
+ {0xab30, 0xab5a, 1},
+ {0xab5c, 0xab5f, 1},
+ {0xab64, 0xab64, 1},
+ {0xfb00, 0xfb06, 1},
+ {0xff21, 0xff3a, 1},
+ {0xff41, 0xff5a, 1},
+ },
+ LatinOffset: 6,
+}
+
+var _Lepcha = &RangeTable{
+ R16: []Range16{
+ {0x1c00, 0x1c37, 1},
+ {0x1c3b, 0x1c49, 1},
+ {0x1c4d, 0x1c4f, 1},
+ },
+}
+
+var _Limbu = &RangeTable{
+ R16: []Range16{
+ {0x1900, 0x191e, 1},
+ {0x1920, 0x192b, 1},
+ {0x1930, 0x193b, 1},
+ {0x1940, 0x1940, 1},
+ {0x1944, 0x194f, 1},
+ },
+}
+
+var _Linear_A = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10600, 0x10736, 1},
+ {0x10740, 0x10755, 1},
+ {0x10760, 0x10767, 1},
+ },
+}
+
+var _Linear_B = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10000, 0x1000b, 1},
+ {0x1000d, 0x10026, 1},
+ {0x10028, 0x1003a, 1},
+ {0x1003c, 0x1003d, 1},
+ {0x1003f, 0x1004d, 1},
+ {0x10050, 0x1005d, 1},
+ {0x10080, 0x100fa, 1},
+ },
+}
+
+var _Lisu = &RangeTable{
+ R16: []Range16{
+ {0xa4d0, 0xa4ff, 1},
+ },
+}
+
+var _Lycian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10280, 0x1029c, 1},
+ },
+}
+
+var _Lydian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10920, 0x10939, 1},
+ {0x1093f, 0x1093f, 1},
+ },
+}
+
+var _Mahajani = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11150, 0x11176, 1},
+ },
+}
+
+var _Malayalam = &RangeTable{
+ R16: []Range16{
+ {0x0d01, 0x0d03, 1},
+ {0x0d05, 0x0d0c, 1},
+ {0x0d0e, 0x0d10, 1},
+ {0x0d12, 0x0d3a, 1},
+ {0x0d3d, 0x0d44, 1},
+ {0x0d46, 0x0d48, 1},
+ {0x0d4a, 0x0d4e, 1},
+ {0x0d57, 0x0d57, 1},
+ {0x0d60, 0x0d63, 1},
+ {0x0d66, 0x0d75, 1},
+ {0x0d79, 0x0d7f, 1},
+ },
+}
+
+var _Mandaic = &RangeTable{
+ R16: []Range16{
+ {0x0840, 0x085b, 1},
+ {0x085e, 0x085e, 1},
+ },
+}
+
+var _Manichaean = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10ac0, 0x10ae6, 1},
+ {0x10aeb, 0x10af6, 1},
+ },
+}
+
+var _Meetei_Mayek = &RangeTable{
+ R16: []Range16{
+ {0xaae0, 0xaaf6, 1},
+ {0xabc0, 0xabed, 1},
+ {0xabf0, 0xabf9, 1},
+ },
+}
+
+var _Mende_Kikakui = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x1e800, 0x1e8c4, 1},
+ {0x1e8c7, 0x1e8d6, 1},
+ },
+}
+
+var _Meroitic_Cursive = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x109a0, 0x109b7, 1},
+ {0x109be, 0x109bf, 1},
+ },
+}
+
+var _Meroitic_Hieroglyphs = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10980, 0x1099f, 1},
+ },
+}
+
+var _Miao = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x16f00, 0x16f44, 1},
+ {0x16f50, 0x16f7e, 1},
+ {0x16f8f, 0x16f9f, 1},
+ },
+}
+
+var _Modi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11600, 0x11644, 1},
+ {0x11650, 0x11659, 1},
+ },
+}
+
+var _Mongolian = &RangeTable{
+ R16: []Range16{
+ {0x1800, 0x1801, 1},
+ {0x1804, 0x1804, 1},
+ {0x1806, 0x180e, 1},
+ {0x1810, 0x1819, 1},
+ {0x1820, 0x1877, 1},
+ {0x1880, 0x18aa, 1},
+ },
+}
+
+var _Mro = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x16a40, 0x16a5e, 1},
+ {0x16a60, 0x16a69, 1},
+ {0x16a6e, 0x16a6f, 1},
+ },
+}
+
+var _Myanmar = &RangeTable{
+ R16: []Range16{
+ {0x1000, 0x109f, 1},
+ {0xa9e0, 0xa9fe, 1},
+ {0xaa60, 0xaa7f, 1},
+ },
+}
+
+var _Nabataean = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10880, 0x1089e, 1},
+ {0x108a7, 0x108af, 1},
+ },
+}
+
+var _New_Tai_Lue = &RangeTable{
+ R16: []Range16{
+ {0x1980, 0x19ab, 1},
+ {0x19b0, 0x19c9, 1},
+ {0x19d0, 0x19da, 1},
+ {0x19de, 0x19df, 1},
+ },
+}
+
+var _Nko = &RangeTable{
+ R16: []Range16{
+ {0x07c0, 0x07fa, 1},
+ },
+}
+
+var _Ogham = &RangeTable{
+ R16: []Range16{
+ {0x1680, 0x169c, 1},
+ },
+}
+
+var _Ol_Chiki = &RangeTable{
+ R16: []Range16{
+ {0x1c50, 0x1c7f, 1},
+ },
+}
+
+var _Old_Italic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10300, 0x10323, 1},
+ },
+}
+
+var _Old_North_Arabian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10a80, 0x10a9f, 1},
+ },
+}
+
+var _Old_Permic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10350, 0x1037a, 1},
+ },
+}
+
+var _Old_Persian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x103a0, 0x103c3, 1},
+ {0x103c8, 0x103d5, 1},
+ },
+}
+
+var _Old_South_Arabian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10a60, 0x10a7f, 1},
+ },
+}
+
+var _Old_Turkic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10c00, 0x10c48, 1},
+ },
+}
+
+var _Oriya = &RangeTable{
+ R16: []Range16{
+ {0x0b01, 0x0b03, 1},
+ {0x0b05, 0x0b0c, 1},
+ {0x0b0f, 0x0b10, 1},
+ {0x0b13, 0x0b28, 1},
+ {0x0b2a, 0x0b30, 1},
+ {0x0b32, 0x0b33, 1},
+ {0x0b35, 0x0b39, 1},
+ {0x0b3c, 0x0b44, 1},
+ {0x0b47, 0x0b48, 1},
+ {0x0b4b, 0x0b4d, 1},
+ {0x0b56, 0x0b57, 1},
+ {0x0b5c, 0x0b5d, 1},
+ {0x0b5f, 0x0b63, 1},
+ {0x0b66, 0x0b77, 1},
+ },
+}
+
+var _Osmanya = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10480, 0x1049d, 1},
+ {0x104a0, 0x104a9, 1},
+ },
+}
+
+var _Pahawh_Hmong = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x16b00, 0x16b45, 1},
+ {0x16b50, 0x16b59, 1},
+ {0x16b5b, 0x16b61, 1},
+ {0x16b63, 0x16b77, 1},
+ {0x16b7d, 0x16b8f, 1},
+ },
+}
+
+var _Palmyrene = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10860, 0x1087f, 1},
+ },
+}
+
+var _Pau_Cin_Hau = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11ac0, 0x11af8, 1},
+ },
+}
+
+var _Phags_Pa = &RangeTable{
+ R16: []Range16{
+ {0xa840, 0xa877, 1},
+ },
+}
+
+var _Phoenician = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10900, 0x1091b, 1},
+ {0x1091f, 0x1091f, 1},
+ },
+}
+
+var _Psalter_Pahlavi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10b80, 0x10b91, 1},
+ {0x10b99, 0x10b9c, 1},
+ {0x10ba9, 0x10baf, 1},
+ },
+}
+
+var _Rejang = &RangeTable{
+ R16: []Range16{
+ {0xa930, 0xa953, 1},
+ {0xa95f, 0xa95f, 1},
+ },
+}
+
+var _Runic = &RangeTable{
+ R16: []Range16{
+ {0x16a0, 0x16ea, 1},
+ {0x16ee, 0x16f8, 1},
+ },
+}
+
+var _Samaritan = &RangeTable{
+ R16: []Range16{
+ {0x0800, 0x082d, 1},
+ {0x0830, 0x083e, 1},
+ },
+}
+
+var _Saurashtra = &RangeTable{
+ R16: []Range16{
+ {0xa880, 0xa8c4, 1},
+ {0xa8ce, 0xa8d9, 1},
+ },
+}
+
+var _Sharada = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11180, 0x111c8, 1},
+ {0x111cd, 0x111cd, 1},
+ {0x111d0, 0x111da, 1},
+ },
+}
+
+var _Shavian = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10450, 0x1047f, 1},
+ },
+}
+
+var _Siddham = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11580, 0x115b5, 1},
+ {0x115b8, 0x115c9, 1},
+ },
+}
+
+var _Sinhala = &RangeTable{
+ R16: []Range16{
+ {0x0d82, 0x0d83, 1},
+ {0x0d85, 0x0d96, 1},
+ {0x0d9a, 0x0db1, 1},
+ {0x0db3, 0x0dbb, 1},
+ {0x0dbd, 0x0dbd, 1},
+ {0x0dc0, 0x0dc6, 1},
+ {0x0dca, 0x0dca, 1},
+ {0x0dcf, 0x0dd4, 1},
+ {0x0dd6, 0x0dd6, 1},
+ {0x0dd8, 0x0ddf, 1},
+ {0x0de6, 0x0def, 1},
+ {0x0df2, 0x0df4, 1},
+ },
+ R32: []Range32{
+ {0x111e1, 0x111f4, 1},
+ },
+}
+
+var _Sora_Sompeng = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x110d0, 0x110e8, 1},
+ {0x110f0, 0x110f9, 1},
+ },
+}
+
+var _Sundanese = &RangeTable{
+ R16: []Range16{
+ {0x1b80, 0x1bbf, 1},
+ {0x1cc0, 0x1cc7, 1},
+ },
+}
+
+var _Syloti_Nagri = &RangeTable{
+ R16: []Range16{
+ {0xa800, 0xa82b, 1},
+ },
+}
+
+var _Syriac = &RangeTable{
+ R16: []Range16{
+ {0x0700, 0x070d, 1},
+ {0x070f, 0x074a, 1},
+ {0x074d, 0x074f, 1},
+ },
+}
+
+var _Tagalog = &RangeTable{
+ R16: []Range16{
+ {0x1700, 0x170c, 1},
+ {0x170e, 0x1714, 1},
+ },
+}
+
+var _Tagbanwa = &RangeTable{
+ R16: []Range16{
+ {0x1760, 0x176c, 1},
+ {0x176e, 0x1770, 1},
+ {0x1772, 0x1773, 1},
+ },
+}
+
+var _Tai_Le = &RangeTable{
+ R16: []Range16{
+ {0x1950, 0x196d, 1},
+ {0x1970, 0x1974, 1},
+ },
+}
+
+var _Tai_Tham = &RangeTable{
+ R16: []Range16{
+ {0x1a20, 0x1a5e, 1},
+ {0x1a60, 0x1a7c, 1},
+ {0x1a7f, 0x1a89, 1},
+ {0x1a90, 0x1a99, 1},
+ {0x1aa0, 0x1aad, 1},
+ },
+}
+
+var _Tai_Viet = &RangeTable{
+ R16: []Range16{
+ {0xaa80, 0xaac2, 1},
+ {0xaadb, 0xaadf, 1},
+ },
+}
+
+var _Takri = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11680, 0x116b7, 1},
+ {0x116c0, 0x116c9, 1},
+ },
+}
+
+var _Tamil = &RangeTable{
+ R16: []Range16{
+ {0x0b82, 0x0b83, 1},
+ {0x0b85, 0x0b8a, 1},
+ {0x0b8e, 0x0b90, 1},
+ {0x0b92, 0x0b95, 1},
+ {0x0b99, 0x0b9a, 1},
+ {0x0b9c, 0x0b9c, 1},
+ {0x0b9e, 0x0b9f, 1},
+ {0x0ba3, 0x0ba4, 1},
+ {0x0ba8, 0x0baa, 1},
+ {0x0bae, 0x0bb9, 1},
+ {0x0bbe, 0x0bc2, 1},
+ {0x0bc6, 0x0bc8, 1},
+ {0x0bca, 0x0bcd, 1},
+ {0x0bd0, 0x0bd0, 1},
+ {0x0bd7, 0x0bd7, 1},
+ {0x0be6, 0x0bfa, 1},
+ },
+}
+
+var _Telugu = &RangeTable{
+ R16: []Range16{
+ {0x0c00, 0x0c03, 1},
+ {0x0c05, 0x0c0c, 1},
+ {0x0c0e, 0x0c10, 1},
+ {0x0c12, 0x0c28, 1},
+ {0x0c2a, 0x0c39, 1},
+ {0x0c3d, 0x0c44, 1},
+ {0x0c46, 0x0c48, 1},
+ {0x0c4a, 0x0c4d, 1},
+ {0x0c55, 0x0c56, 1},
+ {0x0c58, 0x0c59, 1},
+ {0x0c60, 0x0c63, 1},
+ {0x0c66, 0x0c6f, 1},
+ {0x0c78, 0x0c7f, 1},
+ },
+}
+
+var _Thaana = &RangeTable{
+ R16: []Range16{
+ {0x0780, 0x07b1, 1},
+ },
+}
+
+var _Thai = &RangeTable{
+ R16: []Range16{
+ {0x0e01, 0x0e3a, 1},
+ {0x0e40, 0x0e5b, 1},
+ },
+}
+
+var _Tibetan = &RangeTable{
+ R16: []Range16{
+ {0x0f00, 0x0f47, 1},
+ {0x0f49, 0x0f6c, 1},
+ {0x0f71, 0x0f97, 1},
+ {0x0f99, 0x0fbc, 1},
+ {0x0fbe, 0x0fcc, 1},
+ {0x0fce, 0x0fd4, 1},
+ {0x0fd9, 0x0fda, 1},
+ },
+}
+
+var _Tifinagh = &RangeTable{
+ R16: []Range16{
+ {0x2d30, 0x2d67, 1},
+ {0x2d6f, 0x2d70, 1},
+ {0x2d7f, 0x2d7f, 1},
+ },
+}
+
+var _Tirhuta = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x11480, 0x114c7, 1},
+ {0x114d0, 0x114d9, 1},
+ },
+}
+
+var _Ugaritic = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x10380, 0x1039d, 1},
+ {0x1039f, 0x1039f, 1},
+ },
+}
+
+var _Vai = &RangeTable{
+ R16: []Range16{
+ {0xa500, 0xa62b, 1},
+ },
+}
+
+var _Warang_Citi = &RangeTable{
+ R16: []Range16{},
+ R32: []Range32{
+ {0x118a0, 0x118f2, 1},
+ {0x118ff, 0x118ff, 1},
+ },
+}
+
+var _Yi = &RangeTable{
+ R16: []Range16{
+ {0xa000, 0xa48c, 1},
+ {0xa490, 0xa4c6, 1},
+ },
+}
+
+// These variables have type *RangeTable.
+var (
+ Arabic = _Arabic // Arabic is the set of Unicode characters in script Arabic.
+ Armenian = _Armenian // Armenian is the set of Unicode characters in script Armenian.
+ Avestan = _Avestan // Avestan is the set of Unicode characters in script Avestan.
+ Balinese = _Balinese // Balinese is the set of Unicode characters in script Balinese.
+ Bamum = _Bamum // Bamum is the set of Unicode characters in script Bamum.
+ Bassa_Vah = _Bassa_Vah // Bassa_Vah is the set of Unicode characters in script Bassa_Vah.
+ Batak = _Batak // Batak is the set of Unicode characters in script Batak.
+ Bengali = _Bengali // Bengali is the set of Unicode characters in script Bengali.
+ Bopomofo = _Bopomofo // Bopomofo is the set of Unicode characters in script Bopomofo.
+ Brahmi = _Brahmi // Brahmi is the set of Unicode characters in script Brahmi.
+ Braille = _Braille // Braille is the set of Unicode characters in script Braille.
+ Buginese = _Buginese // Buginese is the set of Unicode characters in script Buginese.
+ Buhid = _Buhid // Buhid is the set of Unicode characters in script Buhid.
+ Canadian_Aboriginal = _Canadian_Aboriginal // Canadian_Aboriginal is the set of Unicode characters in script Canadian_Aboriginal.
+ Carian = _Carian // Carian is the set of Unicode characters in script Carian.
+ Caucasian_Albanian = _Caucasian_Albanian // Caucasian_Albanian is the set of Unicode characters in script Caucasian_Albanian.
+ Chakma = _Chakma // Chakma is the set of Unicode characters in script Chakma.
+ Cham = _Cham // Cham is the set of Unicode characters in script Cham.
+ Cherokee = _Cherokee // Cherokee is the set of Unicode characters in script Cherokee.
+ Common = _Common // Common is the set of Unicode characters in script Common.
+ Coptic = _Coptic // Coptic is the set of Unicode characters in script Coptic.
+ Cuneiform = _Cuneiform // Cuneiform is the set of Unicode characters in script Cuneiform.
+ Cypriot = _Cypriot // Cypriot is the set of Unicode characters in script Cypriot.
+ Cyrillic = _Cyrillic // Cyrillic is the set of Unicode characters in script Cyrillic.
+ Deseret = _Deseret // Deseret is the set of Unicode characters in script Deseret.
+ Devanagari = _Devanagari // Devanagari is the set of Unicode characters in script Devanagari.
+ Duployan = _Duployan // Duployan is the set of Unicode characters in script Duployan.
+ Egyptian_Hieroglyphs = _Egyptian_Hieroglyphs // Egyptian_Hieroglyphs is the set of Unicode characters in script Egyptian_Hieroglyphs.
+ Elbasan = _Elbasan // Elbasan is the set of Unicode characters in script Elbasan.
+ Ethiopic = _Ethiopic // Ethiopic is the set of Unicode characters in script Ethiopic.
+ Georgian = _Georgian // Georgian is the set of Unicode characters in script Georgian.
+ Glagolitic = _Glagolitic // Glagolitic is the set of Unicode characters in script Glagolitic.
+ Gothic = _Gothic // Gothic is the set of Unicode characters in script Gothic.
+ Grantha = _Grantha // Grantha is the set of Unicode characters in script Grantha.
+ Greek = _Greek // Greek is the set of Unicode characters in script Greek.
+ Gujarati = _Gujarati // Gujarati is the set of Unicode characters in script Gujarati.
+ Gurmukhi = _Gurmukhi // Gurmukhi is the set of Unicode characters in script Gurmukhi.
+ Han = _Han // Han is the set of Unicode characters in script Han.
+ Hangul = _Hangul // Hangul is the set of Unicode characters in script Hangul.
+ Hanunoo = _Hanunoo // Hanunoo is the set of Unicode characters in script Hanunoo.
+ Hebrew = _Hebrew // Hebrew is the set of Unicode characters in script Hebrew.
+ Hiragana = _Hiragana // Hiragana is the set of Unicode characters in script Hiragana.
+ Imperial_Aramaic = _Imperial_Aramaic // Imperial_Aramaic is the set of Unicode characters in script Imperial_Aramaic.
+ Inherited = _Inherited // Inherited is the set of Unicode characters in script Inherited.
+ Inscriptional_Pahlavi = _Inscriptional_Pahlavi // Inscriptional_Pahlavi is the set of Unicode characters in script Inscriptional_Pahlavi.
+ Inscriptional_Parthian = _Inscriptional_Parthian // Inscriptional_Parthian is the set of Unicode characters in script Inscriptional_Parthian.
+ Javanese = _Javanese // Javanese is the set of Unicode characters in script Javanese.
+ Kaithi = _Kaithi // Kaithi is the set of Unicode characters in script Kaithi.
+ Kannada = _Kannada // Kannada is the set of Unicode characters in script Kannada.
+ Katakana = _Katakana // Katakana is the set of Unicode characters in script Katakana.
+ Kayah_Li = _Kayah_Li // Kayah_Li is the set of Unicode characters in script Kayah_Li.
+ Kharoshthi = _Kharoshthi // Kharoshthi is the set of Unicode characters in script Kharoshthi.
+ Khmer = _Khmer // Khmer is the set of Unicode characters in script Khmer.
+ Khojki = _Khojki // Khojki is the set of Unicode characters in script Khojki.
+ Khudawadi = _Khudawadi // Khudawadi is the set of Unicode characters in script Khudawadi.
+ Lao = _Lao // Lao is the set of Unicode characters in script Lao.
+ Latin = _Latin // Latin is the set of Unicode characters in script Latin.
+ Lepcha = _Lepcha // Lepcha is the set of Unicode characters in script Lepcha.
+ Limbu = _Limbu // Limbu is the set of Unicode characters in script Limbu.
+ Linear_A = _Linear_A // Linear_A is the set of Unicode characters in script Linear_A.
+ Linear_B = _Linear_B // Linear_B is the set of Unicode characters in script Linear_B.
+ Lisu = _Lisu // Lisu is the set of Unicode characters in script Lisu.
+ Lycian = _Lycian // Lycian is the set of Unicode characters in script Lycian.
+ Lydian = _Lydian // Lydian is the set of Unicode characters in script Lydian.
+ Mahajani = _Mahajani // Mahajani is the set of Unicode characters in script Mahajani.
+ Malayalam = _Malayalam // Malayalam is the set of Unicode characters in script Malayalam.
+ Mandaic = _Mandaic // Mandaic is the set of Unicode characters in script Mandaic.
+ Manichaean = _Manichaean // Manichaean is the set of Unicode characters in script Manichaean.
+ Meetei_Mayek = _Meetei_Mayek // Meetei_Mayek is the set of Unicode characters in script Meetei_Mayek.
+ Mende_Kikakui = _Mende_Kikakui // Mende_Kikakui is the set of Unicode characters in script Mende_Kikakui.
+ Meroitic_Cursive = _Meroitic_Cursive // Meroitic_Cursive is the set of Unicode characters in script Meroitic_Cursive.
+ Meroitic_Hieroglyphs = _Meroitic_Hieroglyphs // Meroitic_Hieroglyphs is the set of Unicode characters in script Meroitic_Hieroglyphs.
+ Miao = _Miao // Miao is the set of Unicode characters in script Miao.
+ Modi = _Modi // Modi is the set of Unicode characters in script Modi.
+ Mongolian = _Mongolian // Mongolian is the set of Unicode characters in script Mongolian.
+ Mro = _Mro // Mro is the set of Unicode characters in script Mro.
+ Myanmar = _Myanmar // Myanmar is the set of Unicode characters in script Myanmar.
+ Nabataean = _Nabataean // Nabataean is the set of Unicode characters in script Nabataean.
+ New_Tai_Lue = _New_Tai_Lue // New_Tai_Lue is the set of Unicode characters in script New_Tai_Lue.
+ Nko = _Nko // Nko is the set of Unicode characters in script Nko.
+ Ogham = _Ogham // Ogham is the set of Unicode characters in script Ogham.
+ Ol_Chiki = _Ol_Chiki // Ol_Chiki is the set of Unicode characters in script Ol_Chiki.
+ Old_Italic = _Old_Italic // Old_Italic is the set of Unicode characters in script Old_Italic.
+ Old_North_Arabian = _Old_North_Arabian // Old_North_Arabian is the set of Unicode characters in script Old_North_Arabian.
+ Old_Permic = _Old_Permic // Old_Permic is the set of Unicode characters in script Old_Permic.
+ Old_Persian = _Old_Persian // Old_Persian is the set of Unicode characters in script Old_Persian.
+ Old_South_Arabian = _Old_South_Arabian // Old_South_Arabian is the set of Unicode characters in script Old_South_Arabian.
+ Old_Turkic = _Old_Turkic // Old_Turkic is the set of Unicode characters in script Old_Turkic.
+ Oriya = _Oriya // Oriya is the set of Unicode characters in script Oriya.
+ Osmanya = _Osmanya // Osmanya is the set of Unicode characters in script Osmanya.
+ Pahawh_Hmong = _Pahawh_Hmong // Pahawh_Hmong is the set of Unicode characters in script Pahawh_Hmong.
+ Palmyrene = _Palmyrene // Palmyrene is the set of Unicode characters in script Palmyrene.
+ Pau_Cin_Hau = _Pau_Cin_Hau // Pau_Cin_Hau is the set of Unicode characters in script Pau_Cin_Hau.
+ Phags_Pa = _Phags_Pa // Phags_Pa is the set of Unicode characters in script Phags_Pa.
+ Phoenician = _Phoenician // Phoenician is the set of Unicode characters in script Phoenician.
+ Psalter_Pahlavi = _Psalter_Pahlavi // Psalter_Pahlavi is the set of Unicode characters in script Psalter_Pahlavi.
+ Rejang = _Rejang // Rejang is the set of Unicode characters in script Rejang.
+ Runic = _Runic // Runic is the set of Unicode characters in script Runic.
+ Samaritan = _Samaritan // Samaritan is the set of Unicode characters in script Samaritan.
+ Saurashtra = _Saurashtra // Saurashtra is the set of Unicode characters in script Saurashtra.
+ Sharada = _Sharada // Sharada is the set of Unicode characters in script Sharada.
+ Shavian = _Shavian // Shavian is the set of Unicode characters in script Shavian.
+ Siddham = _Siddham // Siddham is the set of Unicode characters in script Siddham.
+ Sinhala = _Sinhala // Sinhala is the set of Unicode characters in script Sinhala.
+ Sora_Sompeng = _Sora_Sompeng // Sora_Sompeng is the set of Unicode characters in script Sora_Sompeng.
+ Sundanese = _Sundanese // Sundanese is the set of Unicode characters in script Sundanese.
+ Syloti_Nagri = _Syloti_Nagri // Syloti_Nagri is the set of Unicode characters in script Syloti_Nagri.
+ Syriac = _Syriac // Syriac is the set of Unicode characters in script Syriac.
+ Tagalog = _Tagalog // Tagalog is the set of Unicode characters in script Tagalog.
+ Tagbanwa = _Tagbanwa // Tagbanwa is the set of Unicode characters in script Tagbanwa.
+ Tai_Le = _Tai_Le // Tai_Le is the set of Unicode characters in script Tai_Le.
+ Tai_Tham = _Tai_Tham // Tai_Tham is the set of Unicode characters in script Tai_Tham.
+ Tai_Viet = _Tai_Viet // Tai_Viet is the set of Unicode characters in script Tai_Viet.
+ Takri = _Takri // Takri is the set of Unicode characters in script Takri.
+ Tamil = _Tamil // Tamil is the set of Unicode characters in script Tamil.
+ Telugu = _Telugu // Telugu is the set of Unicode characters in script Telugu.
+ Thaana = _Thaana // Thaana is the set of Unicode characters in script Thaana.
+ Thai = _Thai // Thai is the set of Unicode characters in script Thai.
+ Tibetan = _Tibetan // Tibetan is the set of Unicode characters in script Tibetan.
+ Tifinagh = _Tifinagh // Tifinagh is the set of Unicode characters in script Tifinagh.
+ Tirhuta = _Tirhuta // Tirhuta is the set of Unicode characters in script Tirhuta.
+ Ugaritic = _Ugaritic // Ugaritic is the set of Unicode characters in script Ugaritic.
+ Vai = _Vai // Vai is the set of Unicode characters in script Vai.
+ Warang_Citi = _Warang_Citi // Warang_Citi is the set of Unicode characters in script Warang_Citi.
+ Yi = _Yi // Yi is the set of Unicode characters in script Yi.
+)
+
+// Generated by running
+// maketables --props=all --url=http://www.unicode.org/Public/7.0.0/ucd/
+// DO NOT EDIT
+
+// Properties is the set of Unicode property tables.
+var Properties = map[string]*RangeTable{
+ "ASCII_Hex_Digit": ASCII_Hex_Digit,
+ "Bidi_Control": Bidi_Control,
+ "Dash": Dash,
+ "Deprecated": Deprecated,
+ "Diacritic": Diacritic,
+ "Extender": Extender,
+ "Hex_Digit": Hex_Digit,
+ "Hyphen": Hyphen,
+ "IDS_Binary_Operator": IDS_Binary_Operator,
+ "IDS_Trinary_Operator": IDS_Trinary_Operator,
+ "Ideographic": Ideographic,
+ "Join_Control": Join_Control,
+ "Logical_Order_Exception": Logical_Order_Exception,
+ "Noncharacter_Code_Point": Noncharacter_Code_Point,
+ "Other_Alphabetic": Other_Alphabetic,
+ "Other_Default_Ignorable_Code_Point": Other_Default_Ignorable_Code_Point,
+ "Other_Grapheme_Extend": Other_Grapheme_Extend,
+ "Other_ID_Continue": Other_ID_Continue,
+ "Other_ID_Start": Other_ID_Start,
+ "Other_Lowercase": Other_Lowercase,
+ "Other_Math": Other_Math,
+ "Other_Uppercase": Other_Uppercase,
+ "Pattern_Syntax": Pattern_Syntax,
+ "Pattern_White_Space": Pattern_White_Space,
+ "Quotation_Mark": Quotation_Mark,
+ "Radical": Radical,
+ "STerm": STerm,
+ "Soft_Dotted": Soft_Dotted,
+ "Terminal_Punctuation": Terminal_Punctuation,
+ "Unified_Ideograph": Unified_Ideograph,
+ "Variation_Selector": Variation_Selector,
+ "White_Space": White_Space,
+}
+
+var _ASCII_Hex_Digit = &RangeTable{
+ R16: []Range16{
+ {0x0030, 0x0039, 1},
+ {0x0041, 0x0046, 1},
+ {0x0061, 0x0066, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _Bidi_Control = &RangeTable{
+ R16: []Range16{
+ {0x061c, 0x061c, 1},
+ {0x200e, 0x200f, 1},
+ {0x202a, 0x202e, 1},
+ {0x2066, 0x2069, 1},
+ },
+}
+
+var _Dash = &RangeTable{
+ R16: []Range16{
+ {0x002d, 0x002d, 1},
+ {0x058a, 0x058a, 1},
+ {0x05be, 0x05be, 1},
+ {0x1400, 0x1400, 1},
+ {0x1806, 0x1806, 1},
+ {0x2010, 0x2015, 1},
+ {0x2053, 0x2053, 1},
+ {0x207b, 0x207b, 1},
+ {0x208b, 0x208b, 1},
+ {0x2212, 0x2212, 1},
+ {0x2e17, 0x2e17, 1},
+ {0x2e1a, 0x2e1a, 1},
+ {0x2e3a, 0x2e3b, 1},
+ {0x2e40, 0x2e40, 1},
+ {0x301c, 0x301c, 1},
+ {0x3030, 0x3030, 1},
+ {0x30a0, 0x30a0, 1},
+ {0xfe31, 0xfe32, 1},
+ {0xfe58, 0xfe58, 1},
+ {0xfe63, 0xfe63, 1},
+ {0xff0d, 0xff0d, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Deprecated = &RangeTable{
+ R16: []Range16{
+ {0x0149, 0x0149, 1},
+ {0x0673, 0x0673, 1},
+ {0x0f77, 0x0f77, 1},
+ {0x0f79, 0x0f79, 1},
+ {0x17a3, 0x17a4, 1},
+ {0x206a, 0x206f, 1},
+ {0x2329, 0x232a, 1},
+ },
+ R32: []Range32{
+ {0xe0001, 0xe0001, 1},
+ {0xe0020, 0xe007f, 1},
+ },
+}
+
+var _Diacritic = &RangeTable{
+ R16: []Range16{
+ {0x005e, 0x005e, 1},
+ {0x0060, 0x0060, 1},
+ {0x00a8, 0x00a8, 1},
+ {0x00af, 0x00af, 1},
+ {0x00b4, 0x00b4, 1},
+ {0x00b7, 0x00b8, 1},
+ {0x02b0, 0x034e, 1},
+ {0x0350, 0x0357, 1},
+ {0x035d, 0x0362, 1},
+ {0x0374, 0x0375, 1},
+ {0x037a, 0x037a, 1},
+ {0x0384, 0x0385, 1},
+ {0x0483, 0x0487, 1},
+ {0x0559, 0x0559, 1},
+ {0x0591, 0x05a1, 1},
+ {0x05a3, 0x05bd, 1},
+ {0x05bf, 0x05bf, 1},
+ {0x05c1, 0x05c2, 1},
+ {0x05c4, 0x05c4, 1},
+ {0x064b, 0x0652, 1},
+ {0x0657, 0x0658, 1},
+ {0x06df, 0x06e0, 1},
+ {0x06e5, 0x06e6, 1},
+ {0x06ea, 0x06ec, 1},
+ {0x0730, 0x074a, 1},
+ {0x07a6, 0x07b0, 1},
+ {0x07eb, 0x07f5, 1},
+ {0x0818, 0x0819, 1},
+ {0x08e4, 0x08fe, 1},
+ {0x093c, 0x093c, 1},
+ {0x094d, 0x094d, 1},
+ {0x0951, 0x0954, 1},
+ {0x0971, 0x0971, 1},
+ {0x09bc, 0x09bc, 1},
+ {0x09cd, 0x09cd, 1},
+ {0x0a3c, 0x0a3c, 1},
+ {0x0a4d, 0x0a4d, 1},
+ {0x0abc, 0x0abc, 1},
+ {0x0acd, 0x0acd, 1},
+ {0x0b3c, 0x0b3c, 1},
+ {0x0b4d, 0x0b4d, 1},
+ {0x0bcd, 0x0bcd, 1},
+ {0x0c4d, 0x0c4d, 1},
+ {0x0cbc, 0x0cbc, 1},
+ {0x0ccd, 0x0ccd, 1},
+ {0x0d4d, 0x0d4d, 1},
+ {0x0dca, 0x0dca, 1},
+ {0x0e47, 0x0e4c, 1},
+ {0x0e4e, 0x0e4e, 1},
+ {0x0ec8, 0x0ecc, 1},
+ {0x0f18, 0x0f19, 1},
+ {0x0f35, 0x0f35, 1},
+ {0x0f37, 0x0f37, 1},
+ {0x0f39, 0x0f39, 1},
+ {0x0f3e, 0x0f3f, 1},
+ {0x0f82, 0x0f84, 1},
+ {0x0f86, 0x0f87, 1},
+ {0x0fc6, 0x0fc6, 1},
+ {0x1037, 0x1037, 1},
+ {0x1039, 0x103a, 1},
+ {0x1087, 0x108d, 1},
+ {0x108f, 0x108f, 1},
+ {0x109a, 0x109b, 1},
+ {0x17c9, 0x17d3, 1},
+ {0x17dd, 0x17dd, 1},
+ {0x1939, 0x193b, 1},
+ {0x1a75, 0x1a7c, 1},
+ {0x1a7f, 0x1a7f, 1},
+ {0x1ab0, 0x1abd, 1},
+ {0x1b34, 0x1b34, 1},
+ {0x1b44, 0x1b44, 1},
+ {0x1b6b, 0x1b73, 1},
+ {0x1baa, 0x1bab, 1},
+ {0x1c36, 0x1c37, 1},
+ {0x1c78, 0x1c7d, 1},
+ {0x1cd0, 0x1ce8, 1},
+ {0x1ced, 0x1ced, 1},
+ {0x1cf4, 0x1cf4, 1},
+ {0x1cf8, 0x1cf9, 1},
+ {0x1d2c, 0x1d6a, 1},
+ {0x1dc4, 0x1dcf, 1},
+ {0x1df5, 0x1df5, 1},
+ {0x1dfd, 0x1dff, 1},
+ {0x1fbd, 0x1fbd, 1},
+ {0x1fbf, 0x1fc1, 1},
+ {0x1fcd, 0x1fcf, 1},
+ {0x1fdd, 0x1fdf, 1},
+ {0x1fed, 0x1fef, 1},
+ {0x1ffd, 0x1ffe, 1},
+ {0x2cef, 0x2cf1, 1},
+ {0x2e2f, 0x2e2f, 1},
+ {0x302a, 0x302f, 1},
+ {0x3099, 0x309c, 1},
+ {0x30fc, 0x30fc, 1},
+ {0xa66f, 0xa66f, 1},
+ {0xa67c, 0xa67d, 1},
+ {0xa67f, 0xa67f, 1},
+ {0xa69c, 0xa69d, 1},
+ {0xa6f0, 0xa6f1, 1},
+ {0xa717, 0xa721, 1},
+ {0xa788, 0xa788, 1},
+ {0xa7f8, 0xa7f9, 1},
+ {0xa8c4, 0xa8c4, 1},
+ {0xa8e0, 0xa8f1, 1},
+ {0xa92b, 0xa92e, 1},
+ {0xa953, 0xa953, 1},
+ {0xa9b3, 0xa9b3, 1},
+ {0xa9c0, 0xa9c0, 1},
+ {0xa9e5, 0xa9e5, 1},
+ {0xaa7b, 0xaa7d, 1},
+ {0xaabf, 0xaac2, 1},
+ {0xaaf6, 0xaaf6, 1},
+ {0xab5b, 0xab5f, 1},
+ {0xabec, 0xabed, 1},
+ {0xfb1e, 0xfb1e, 1},
+ {0xfe20, 0xfe2d, 1},
+ {0xff3e, 0xff3e, 1},
+ {0xff40, 0xff40, 1},
+ {0xff70, 0xff70, 1},
+ {0xff9e, 0xff9f, 1},
+ {0xffe3, 0xffe3, 1},
+ },
+ R32: []Range32{
+ {0x102e0, 0x102e0, 1},
+ {0x10ae5, 0x10ae6, 1},
+ {0x110b9, 0x110ba, 1},
+ {0x11133, 0x11134, 1},
+ {0x11173, 0x11173, 1},
+ {0x111c0, 0x111c0, 1},
+ {0x11235, 0x11236, 1},
+ {0x112e9, 0x112ea, 1},
+ {0x1133c, 0x1133c, 1},
+ {0x1134d, 0x1134d, 1},
+ {0x11366, 0x1136c, 1},
+ {0x11370, 0x11374, 1},
+ {0x114c2, 0x114c3, 1},
+ {0x115bf, 0x115c0, 1},
+ {0x1163f, 0x1163f, 1},
+ {0x116b6, 0x116b7, 1},
+ {0x16af0, 0x16af4, 1},
+ {0x16f8f, 0x16f9f, 1},
+ {0x1d167, 0x1d169, 1},
+ {0x1d16d, 0x1d172, 1},
+ {0x1d17b, 0x1d182, 1},
+ {0x1d185, 0x1d18b, 1},
+ {0x1d1aa, 0x1d1ad, 1},
+ {0x1e8d0, 0x1e8d6, 1},
+ },
+ LatinOffset: 6,
+}
+
+var _Extender = &RangeTable{
+ R16: []Range16{
+ {0x00b7, 0x00b7, 1},
+ {0x02d0, 0x02d1, 1},
+ {0x0640, 0x0640, 1},
+ {0x07fa, 0x07fa, 1},
+ {0x0e46, 0x0e46, 1},
+ {0x0ec6, 0x0ec6, 1},
+ {0x180a, 0x180a, 1},
+ {0x1843, 0x1843, 1},
+ {0x1aa7, 0x1aa7, 1},
+ {0x1c36, 0x1c36, 1},
+ {0x1c7b, 0x1c7b, 1},
+ {0x3005, 0x3005, 1},
+ {0x3031, 0x3035, 1},
+ {0x309d, 0x309e, 1},
+ {0x30fc, 0x30fe, 1},
+ {0xa015, 0xa015, 1},
+ {0xa60c, 0xa60c, 1},
+ {0xa9cf, 0xa9cf, 1},
+ {0xa9e6, 0xa9e6, 1},
+ {0xaa70, 0xaa70, 1},
+ {0xaadd, 0xaadd, 1},
+ {0xaaf3, 0xaaf4, 1},
+ {0xff70, 0xff70, 1},
+ },
+ R32: []Range32{
+ {0x1135d, 0x1135d, 1},
+ {0x115c6, 0x115c8, 1},
+ {0x16b42, 0x16b43, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Hex_Digit = &RangeTable{
+ R16: []Range16{
+ {0x0030, 0x0039, 1},
+ {0x0041, 0x0046, 1},
+ {0x0061, 0x0066, 1},
+ {0xff10, 0xff19, 1},
+ {0xff21, 0xff26, 1},
+ {0xff41, 0xff46, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _Hyphen = &RangeTable{
+ R16: []Range16{
+ {0x002d, 0x002d, 1},
+ {0x00ad, 0x00ad, 1},
+ {0x058a, 0x058a, 1},
+ {0x1806, 0x1806, 1},
+ {0x2010, 0x2011, 1},
+ {0x2e17, 0x2e17, 1},
+ {0x30fb, 0x30fb, 1},
+ {0xfe63, 0xfe63, 1},
+ {0xff0d, 0xff0d, 1},
+ {0xff65, 0xff65, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _IDS_Binary_Operator = &RangeTable{
+ R16: []Range16{
+ {0x2ff0, 0x2ff1, 1},
+ {0x2ff4, 0x2ffb, 1},
+ },
+}
+
+var _IDS_Trinary_Operator = &RangeTable{
+ R16: []Range16{
+ {0x2ff2, 0x2ff3, 1},
+ },
+}
+
+var _Ideographic = &RangeTable{
+ R16: []Range16{
+ {0x3006, 0x3007, 1},
+ {0x3021, 0x3029, 1},
+ {0x3038, 0x303a, 1},
+ {0x3400, 0x4db5, 1},
+ {0x4e00, 0x9fcc, 1},
+ {0xf900, 0xfa6d, 1},
+ {0xfa70, 0xfad9, 1},
+ },
+ R32: []Range32{
+ {0x20000, 0x2a6d6, 1},
+ {0x2a700, 0x2b734, 1},
+ {0x2b740, 0x2b81d, 1},
+ {0x2f800, 0x2fa1d, 1},
+ },
+}
+
+var _Join_Control = &RangeTable{
+ R16: []Range16{
+ {0x200c, 0x200d, 1},
+ },
+}
+
+var _Logical_Order_Exception = &RangeTable{
+ R16: []Range16{
+ {0x0e40, 0x0e44, 1},
+ {0x0ec0, 0x0ec4, 1},
+ {0xaab5, 0xaab6, 1},
+ {0xaab9, 0xaab9, 1},
+ {0xaabb, 0xaabc, 1},
+ },
+}
+
+var _Noncharacter_Code_Point = &RangeTable{
+ R16: []Range16{
+ {0xfdd0, 0xfdef, 1},
+ {0xfffe, 0xffff, 1},
+ },
+ R32: []Range32{
+ {0x1fffe, 0x1ffff, 1},
+ {0x2fffe, 0x2ffff, 1},
+ {0x3fffe, 0x3ffff, 1},
+ {0x4fffe, 0x4ffff, 1},
+ {0x5fffe, 0x5ffff, 1},
+ {0x6fffe, 0x6ffff, 1},
+ {0x7fffe, 0x7ffff, 1},
+ {0x8fffe, 0x8ffff, 1},
+ {0x9fffe, 0x9ffff, 1},
+ {0xafffe, 0xaffff, 1},
+ {0xbfffe, 0xbffff, 1},
+ {0xcfffe, 0xcffff, 1},
+ {0xdfffe, 0xdffff, 1},
+ {0xefffe, 0xeffff, 1},
+ {0xffffe, 0xfffff, 1},
+ {0x10fffe, 0x10ffff, 1},
+ },
+}
+
+var _Other_Alphabetic = &RangeTable{
+ R16: []Range16{
+ {0x0345, 0x0345, 1},
+ {0x05b0, 0x05bd, 1},
+ {0x05bf, 0x05bf, 1},
+ {0x05c1, 0x05c2, 1},
+ {0x05c4, 0x05c5, 1},
+ {0x05c7, 0x05c7, 1},
+ {0x0610, 0x061a, 1},
+ {0x064b, 0x0657, 1},
+ {0x0659, 0x065f, 1},
+ {0x0670, 0x0670, 1},
+ {0x06d6, 0x06dc, 1},
+ {0x06e1, 0x06e4, 1},
+ {0x06e7, 0x06e8, 1},
+ {0x06ed, 0x06ed, 1},
+ {0x0711, 0x0711, 1},
+ {0x0730, 0x073f, 1},
+ {0x07a6, 0x07b0, 1},
+ {0x0816, 0x0817, 1},
+ {0x081b, 0x0823, 1},
+ {0x0825, 0x0827, 1},
+ {0x0829, 0x082c, 1},
+ {0x08e4, 0x08e9, 1},
+ {0x08f0, 0x0903, 1},
+ {0x093a, 0x093b, 1},
+ {0x093e, 0x094c, 1},
+ {0x094e, 0x094f, 1},
+ {0x0955, 0x0957, 1},
+ {0x0962, 0x0963, 1},
+ {0x0981, 0x0983, 1},
+ {0x09be, 0x09c4, 1},
+ {0x09c7, 0x09c8, 1},
+ {0x09cb, 0x09cc, 1},
+ {0x09d7, 0x09d7, 1},
+ {0x09e2, 0x09e3, 1},
+ {0x0a01, 0x0a03, 1},
+ {0x0a3e, 0x0a42, 1},
+ {0x0a47, 0x0a48, 1},
+ {0x0a4b, 0x0a4c, 1},
+ {0x0a51, 0x0a51, 1},
+ {0x0a70, 0x0a71, 1},
+ {0x0a75, 0x0a75, 1},
+ {0x0a81, 0x0a83, 1},
+ {0x0abe, 0x0ac5, 1},
+ {0x0ac7, 0x0ac9, 1},
+ {0x0acb, 0x0acc, 1},
+ {0x0ae2, 0x0ae3, 1},
+ {0x0b01, 0x0b03, 1},
+ {0x0b3e, 0x0b44, 1},
+ {0x0b47, 0x0b48, 1},
+ {0x0b4b, 0x0b4c, 1},
+ {0x0b56, 0x0b57, 1},
+ {0x0b62, 0x0b63, 1},
+ {0x0b82, 0x0b82, 1},
+ {0x0bbe, 0x0bc2, 1},
+ {0x0bc6, 0x0bc8, 1},
+ {0x0bca, 0x0bcc, 1},
+ {0x0bd7, 0x0bd7, 1},
+ {0x0c00, 0x0c03, 1},
+ {0x0c3e, 0x0c44, 1},
+ {0x0c46, 0x0c48, 1},
+ {0x0c4a, 0x0c4c, 1},
+ {0x0c55, 0x0c56, 1},
+ {0x0c62, 0x0c63, 1},
+ {0x0c81, 0x0c83, 1},
+ {0x0cbe, 0x0cc4, 1},
+ {0x0cc6, 0x0cc8, 1},
+ {0x0cca, 0x0ccc, 1},
+ {0x0cd5, 0x0cd6, 1},
+ {0x0ce2, 0x0ce3, 1},
+ {0x0d01, 0x0d03, 1},
+ {0x0d3e, 0x0d44, 1},
+ {0x0d46, 0x0d48, 1},
+ {0x0d4a, 0x0d4c, 1},
+ {0x0d57, 0x0d57, 1},
+ {0x0d62, 0x0d63, 1},
+ {0x0d82, 0x0d83, 1},
+ {0x0dcf, 0x0dd4, 1},
+ {0x0dd6, 0x0dd6, 1},
+ {0x0dd8, 0x0ddf, 1},
+ {0x0df2, 0x0df3, 1},
+ {0x0e31, 0x0e31, 1},
+ {0x0e34, 0x0e3a, 1},
+ {0x0e4d, 0x0e4d, 1},
+ {0x0eb1, 0x0eb1, 1},
+ {0x0eb4, 0x0eb9, 1},
+ {0x0ebb, 0x0ebc, 1},
+ {0x0ecd, 0x0ecd, 1},
+ {0x0f71, 0x0f81, 1},
+ {0x0f8d, 0x0f97, 1},
+ {0x0f99, 0x0fbc, 1},
+ {0x102b, 0x1036, 1},
+ {0x1038, 0x1038, 1},
+ {0x103b, 0x103e, 1},
+ {0x1056, 0x1059, 1},
+ {0x105e, 0x1060, 1},
+ {0x1062, 0x1062, 1},
+ {0x1067, 0x1068, 1},
+ {0x1071, 0x1074, 1},
+ {0x1082, 0x1086, 1},
+ {0x109c, 0x109d, 1},
+ {0x135f, 0x135f, 1},
+ {0x1712, 0x1713, 1},
+ {0x1732, 0x1733, 1},
+ {0x1752, 0x1753, 1},
+ {0x1772, 0x1773, 1},
+ {0x17b6, 0x17c8, 1},
+ {0x18a9, 0x18a9, 1},
+ {0x1920, 0x192b, 1},
+ {0x1930, 0x1938, 1},
+ {0x19b0, 0x19c0, 1},
+ {0x19c8, 0x19c9, 1},
+ {0x1a17, 0x1a1b, 1},
+ {0x1a55, 0x1a5e, 1},
+ {0x1a61, 0x1a74, 1},
+ {0x1b00, 0x1b04, 1},
+ {0x1b35, 0x1b43, 1},
+ {0x1b80, 0x1b82, 1},
+ {0x1ba1, 0x1ba9, 1},
+ {0x1bac, 0x1bad, 1},
+ {0x1be7, 0x1bf1, 1},
+ {0x1c24, 0x1c35, 1},
+ {0x1cf2, 0x1cf3, 1},
+ {0x1de7, 0x1df4, 1},
+ {0x24b6, 0x24e9, 1},
+ {0x2de0, 0x2dff, 1},
+ {0xa674, 0xa67b, 1},
+ {0xa69f, 0xa69f, 1},
+ {0xa823, 0xa827, 1},
+ {0xa880, 0xa881, 1},
+ {0xa8b4, 0xa8c3, 1},
+ {0xa926, 0xa92a, 1},
+ {0xa947, 0xa952, 1},
+ {0xa980, 0xa983, 1},
+ {0xa9b4, 0xa9bf, 1},
+ {0xaa29, 0xaa36, 1},
+ {0xaa43, 0xaa43, 1},
+ {0xaa4c, 0xaa4d, 1},
+ {0xaab0, 0xaab0, 1},
+ {0xaab2, 0xaab4, 1},
+ {0xaab7, 0xaab8, 1},
+ {0xaabe, 0xaabe, 1},
+ {0xaaeb, 0xaaef, 1},
+ {0xaaf5, 0xaaf5, 1},
+ {0xabe3, 0xabea, 1},
+ {0xfb1e, 0xfb1e, 1},
+ },
+ R32: []Range32{
+ {0x10376, 0x1037a, 1},
+ {0x10a01, 0x10a03, 1},
+ {0x10a05, 0x10a06, 1},
+ {0x10a0c, 0x10a0f, 1},
+ {0x11000, 0x11002, 1},
+ {0x11038, 0x11045, 1},
+ {0x11082, 0x11082, 1},
+ {0x110b0, 0x110b8, 1},
+ {0x11100, 0x11102, 1},
+ {0x11127, 0x11132, 1},
+ {0x11180, 0x11182, 1},
+ {0x111b3, 0x111bf, 1},
+ {0x1122c, 0x11234, 1},
+ {0x11237, 0x11237, 1},
+ {0x112df, 0x112e8, 1},
+ {0x11301, 0x11303, 1},
+ {0x1133e, 0x11344, 1},
+ {0x11347, 0x11348, 1},
+ {0x1134b, 0x1134c, 1},
+ {0x11357, 0x11357, 1},
+ {0x11362, 0x11363, 1},
+ {0x114b0, 0x114c1, 1},
+ {0x115af, 0x115b5, 1},
+ {0x115b8, 0x115be, 1},
+ {0x11630, 0x1163e, 1},
+ {0x11640, 0x11640, 1},
+ {0x116ab, 0x116b5, 1},
+ {0x16b30, 0x16b36, 1},
+ {0x16f51, 0x16f7e, 1},
+ {0x1bc9e, 0x1bc9e, 1},
+ {0x1f130, 0x1f149, 1},
+ {0x1f150, 0x1f169, 1},
+ {0x1f170, 0x1f189, 1},
+ },
+}
+
+var _Other_Default_Ignorable_Code_Point = &RangeTable{
+ R16: []Range16{
+ {0x034f, 0x034f, 1},
+ {0x115f, 0x1160, 1},
+ {0x17b4, 0x17b5, 1},
+ {0x2065, 0x2065, 1},
+ {0x3164, 0x3164, 1},
+ {0xffa0, 0xffa0, 1},
+ {0xfff0, 0xfff8, 1},
+ },
+ R32: []Range32{
+ {0xe0000, 0xe0000, 1},
+ {0xe0002, 0xe001f, 1},
+ {0xe0080, 0xe00ff, 1},
+ {0xe01f0, 0xe0fff, 1},
+ },
+}
+
+var _Other_Grapheme_Extend = &RangeTable{
+ R16: []Range16{
+ {0x09be, 0x09be, 1},
+ {0x09d7, 0x09d7, 1},
+ {0x0b3e, 0x0b3e, 1},
+ {0x0b57, 0x0b57, 1},
+ {0x0bbe, 0x0bbe, 1},
+ {0x0bd7, 0x0bd7, 1},
+ {0x0cc2, 0x0cc2, 1},
+ {0x0cd5, 0x0cd6, 1},
+ {0x0d3e, 0x0d3e, 1},
+ {0x0d57, 0x0d57, 1},
+ {0x0dcf, 0x0dcf, 1},
+ {0x0ddf, 0x0ddf, 1},
+ {0x200c, 0x200d, 1},
+ {0x302e, 0x302f, 1},
+ {0xff9e, 0xff9f, 1},
+ },
+ R32: []Range32{
+ {0x1133e, 0x1133e, 1},
+ {0x11357, 0x11357, 1},
+ {0x114b0, 0x114b0, 1},
+ {0x114bd, 0x114bd, 1},
+ {0x115af, 0x115af, 1},
+ {0x1d165, 0x1d165, 1},
+ {0x1d16e, 0x1d172, 1},
+ },
+}
+
+var _Other_ID_Continue = &RangeTable{
+ R16: []Range16{
+ {0x00b7, 0x00b7, 1},
+ {0x0387, 0x0387, 1},
+ {0x1369, 0x1371, 1},
+ {0x19da, 0x19da, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Other_ID_Start = &RangeTable{
+ R16: []Range16{
+ {0x2118, 0x2118, 1},
+ {0x212e, 0x212e, 1},
+ {0x309b, 0x309c, 1},
+ },
+}
+
+var _Other_Lowercase = &RangeTable{
+ R16: []Range16{
+ {0x00aa, 0x00aa, 1},
+ {0x00ba, 0x00ba, 1},
+ {0x02b0, 0x02b8, 1},
+ {0x02c0, 0x02c1, 1},
+ {0x02e0, 0x02e4, 1},
+ {0x0345, 0x0345, 1},
+ {0x037a, 0x037a, 1},
+ {0x1d2c, 0x1d6a, 1},
+ {0x1d78, 0x1d78, 1},
+ {0x1d9b, 0x1dbf, 1},
+ {0x2071, 0x2071, 1},
+ {0x207f, 0x207f, 1},
+ {0x2090, 0x209c, 1},
+ {0x2170, 0x217f, 1},
+ {0x24d0, 0x24e9, 1},
+ {0x2c7c, 0x2c7d, 1},
+ {0xa69c, 0xa69d, 1},
+ {0xa770, 0xa770, 1},
+ {0xa7f8, 0xa7f9, 1},
+ {0xab5c, 0xab5f, 1},
+ },
+ LatinOffset: 2,
+}
+
+var _Other_Math = &RangeTable{
+ R16: []Range16{
+ {0x005e, 0x005e, 1},
+ {0x03d0, 0x03d2, 1},
+ {0x03d5, 0x03d5, 1},
+ {0x03f0, 0x03f1, 1},
+ {0x03f4, 0x03f5, 1},
+ {0x2016, 0x2016, 1},
+ {0x2032, 0x2034, 1},
+ {0x2040, 0x2040, 1},
+ {0x2061, 0x2064, 1},
+ {0x207d, 0x207e, 1},
+ {0x208d, 0x208e, 1},
+ {0x20d0, 0x20dc, 1},
+ {0x20e1, 0x20e1, 1},
+ {0x20e5, 0x20e6, 1},
+ {0x20eb, 0x20ef, 1},
+ {0x2102, 0x2102, 1},
+ {0x2107, 0x2107, 1},
+ {0x210a, 0x2113, 1},
+ {0x2115, 0x2115, 1},
+ {0x2119, 0x211d, 1},
+ {0x2124, 0x2124, 1},
+ {0x2128, 0x2129, 1},
+ {0x212c, 0x212d, 1},
+ {0x212f, 0x2131, 1},
+ {0x2133, 0x2138, 1},
+ {0x213c, 0x213f, 1},
+ {0x2145, 0x2149, 1},
+ {0x2195, 0x2199, 1},
+ {0x219c, 0x219f, 1},
+ {0x21a1, 0x21a2, 1},
+ {0x21a4, 0x21a5, 1},
+ {0x21a7, 0x21a7, 1},
+ {0x21a9, 0x21ad, 1},
+ {0x21b0, 0x21b1, 1},
+ {0x21b6, 0x21b7, 1},
+ {0x21bc, 0x21cd, 1},
+ {0x21d0, 0x21d1, 1},
+ {0x21d3, 0x21d3, 1},
+ {0x21d5, 0x21db, 1},
+ {0x21dd, 0x21dd, 1},
+ {0x21e4, 0x21e5, 1},
+ {0x2308, 0x230b, 1},
+ {0x23b4, 0x23b5, 1},
+ {0x23b7, 0x23b7, 1},
+ {0x23d0, 0x23d0, 1},
+ {0x23e2, 0x23e2, 1},
+ {0x25a0, 0x25a1, 1},
+ {0x25ae, 0x25b6, 1},
+ {0x25bc, 0x25c0, 1},
+ {0x25c6, 0x25c7, 1},
+ {0x25ca, 0x25cb, 1},
+ {0x25cf, 0x25d3, 1},
+ {0x25e2, 0x25e2, 1},
+ {0x25e4, 0x25e4, 1},
+ {0x25e7, 0x25ec, 1},
+ {0x2605, 0x2606, 1},
+ {0x2640, 0x2640, 1},
+ {0x2642, 0x2642, 1},
+ {0x2660, 0x2663, 1},
+ {0x266d, 0x266e, 1},
+ {0x27c5, 0x27c6, 1},
+ {0x27e6, 0x27ef, 1},
+ {0x2983, 0x2998, 1},
+ {0x29d8, 0x29db, 1},
+ {0x29fc, 0x29fd, 1},
+ {0xfe61, 0xfe61, 1},
+ {0xfe63, 0xfe63, 1},
+ {0xfe68, 0xfe68, 1},
+ {0xff3c, 0xff3c, 1},
+ {0xff3e, 0xff3e, 1},
+ },
+ R32: []Range32{
+ {0x1d400, 0x1d454, 1},
+ {0x1d456, 0x1d49c, 1},
+ {0x1d49e, 0x1d49f, 1},
+ {0x1d4a2, 0x1d4a2, 1},
+ {0x1d4a5, 0x1d4a6, 1},
+ {0x1d4a9, 0x1d4ac, 1},
+ {0x1d4ae, 0x1d4b9, 1},
+ {0x1d4bb, 0x1d4bb, 1},
+ {0x1d4bd, 0x1d4c3, 1},
+ {0x1d4c5, 0x1d505, 1},
+ {0x1d507, 0x1d50a, 1},
+ {0x1d50d, 0x1d514, 1},
+ {0x1d516, 0x1d51c, 1},
+ {0x1d51e, 0x1d539, 1},
+ {0x1d53b, 0x1d53e, 1},
+ {0x1d540, 0x1d544, 1},
+ {0x1d546, 0x1d546, 1},
+ {0x1d54a, 0x1d550, 1},
+ {0x1d552, 0x1d6a5, 1},
+ {0x1d6a8, 0x1d6c0, 1},
+ {0x1d6c2, 0x1d6da, 1},
+ {0x1d6dc, 0x1d6fa, 1},
+ {0x1d6fc, 0x1d714, 1},
+ {0x1d716, 0x1d734, 1},
+ {0x1d736, 0x1d74e, 1},
+ {0x1d750, 0x1d76e, 1},
+ {0x1d770, 0x1d788, 1},
+ {0x1d78a, 0x1d7a8, 1},
+ {0x1d7aa, 0x1d7c2, 1},
+ {0x1d7c4, 0x1d7cb, 1},
+ {0x1d7ce, 0x1d7ff, 1},
+ {0x1ee00, 0x1ee03, 1},
+ {0x1ee05, 0x1ee1f, 1},
+ {0x1ee21, 0x1ee22, 1},
+ {0x1ee24, 0x1ee24, 1},
+ {0x1ee27, 0x1ee27, 1},
+ {0x1ee29, 0x1ee32, 1},
+ {0x1ee34, 0x1ee37, 1},
+ {0x1ee39, 0x1ee39, 1},
+ {0x1ee3b, 0x1ee3b, 1},
+ {0x1ee42, 0x1ee42, 1},
+ {0x1ee47, 0x1ee47, 1},
+ {0x1ee49, 0x1ee49, 1},
+ {0x1ee4b, 0x1ee4b, 1},
+ {0x1ee4d, 0x1ee4f, 1},
+ {0x1ee51, 0x1ee52, 1},
+ {0x1ee54, 0x1ee54, 1},
+ {0x1ee57, 0x1ee57, 1},
+ {0x1ee59, 0x1ee59, 1},
+ {0x1ee5b, 0x1ee5b, 1},
+ {0x1ee5d, 0x1ee5d, 1},
+ {0x1ee5f, 0x1ee5f, 1},
+ {0x1ee61, 0x1ee62, 1},
+ {0x1ee64, 0x1ee64, 1},
+ {0x1ee67, 0x1ee6a, 1},
+ {0x1ee6c, 0x1ee72, 1},
+ {0x1ee74, 0x1ee77, 1},
+ {0x1ee79, 0x1ee7c, 1},
+ {0x1ee7e, 0x1ee7e, 1},
+ {0x1ee80, 0x1ee89, 1},
+ {0x1ee8b, 0x1ee9b, 1},
+ {0x1eea1, 0x1eea3, 1},
+ {0x1eea5, 0x1eea9, 1},
+ {0x1eeab, 0x1eebb, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Other_Uppercase = &RangeTable{
+ R16: []Range16{
+ {0x2160, 0x216f, 1},
+ {0x24b6, 0x24cf, 1},
+ },
+ R32: []Range32{
+ {0x1f130, 0x1f149, 1},
+ {0x1f150, 0x1f169, 1},
+ {0x1f170, 0x1f189, 1},
+ },
+}
+
+var _Pattern_Syntax = &RangeTable{
+ R16: []Range16{
+ {0x0021, 0x002f, 1},
+ {0x003a, 0x0040, 1},
+ {0x005b, 0x005e, 1},
+ {0x0060, 0x0060, 1},
+ {0x007b, 0x007e, 1},
+ {0x00a1, 0x00a7, 1},
+ {0x00a9, 0x00a9, 1},
+ {0x00ab, 0x00ac, 1},
+ {0x00ae, 0x00ae, 1},
+ {0x00b0, 0x00b1, 1},
+ {0x00b6, 0x00b6, 1},
+ {0x00bb, 0x00bb, 1},
+ {0x00bf, 0x00bf, 1},
+ {0x00d7, 0x00d7, 1},
+ {0x00f7, 0x00f7, 1},
+ {0x2010, 0x2027, 1},
+ {0x2030, 0x203e, 1},
+ {0x2041, 0x2053, 1},
+ {0x2055, 0x205e, 1},
+ {0x2190, 0x245f, 1},
+ {0x2500, 0x2775, 1},
+ {0x2794, 0x2bff, 1},
+ {0x2e00, 0x2e7f, 1},
+ {0x3001, 0x3003, 1},
+ {0x3008, 0x3020, 1},
+ {0x3030, 0x3030, 1},
+ {0xfd3e, 0xfd3f, 1},
+ {0xfe45, 0xfe46, 1},
+ },
+ LatinOffset: 15,
+}
+
+var _Pattern_White_Space = &RangeTable{
+ R16: []Range16{
+ {0x0009, 0x000d, 1},
+ {0x0020, 0x0020, 1},
+ {0x0085, 0x0085, 1},
+ {0x200e, 0x200f, 1},
+ {0x2028, 0x2029, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _Quotation_Mark = &RangeTable{
+ R16: []Range16{
+ {0x0022, 0x0022, 1},
+ {0x0027, 0x0027, 1},
+ {0x00ab, 0x00ab, 1},
+ {0x00bb, 0x00bb, 1},
+ {0x2018, 0x201f, 1},
+ {0x2039, 0x203a, 1},
+ {0x2e42, 0x2e42, 1},
+ {0x300c, 0x300f, 1},
+ {0x301d, 0x301f, 1},
+ {0xfe41, 0xfe44, 1},
+ {0xff02, 0xff02, 1},
+ {0xff07, 0xff07, 1},
+ {0xff62, 0xff63, 1},
+ },
+ LatinOffset: 4,
+}
+
+var _Radical = &RangeTable{
+ R16: []Range16{
+ {0x2e80, 0x2e99, 1},
+ {0x2e9b, 0x2ef3, 1},
+ {0x2f00, 0x2fd5, 1},
+ },
+}
+
+var _STerm = &RangeTable{
+ R16: []Range16{
+ {0x0021, 0x0021, 1},
+ {0x002e, 0x002e, 1},
+ {0x003f, 0x003f, 1},
+ {0x0589, 0x0589, 1},
+ {0x061f, 0x061f, 1},
+ {0x06d4, 0x06d4, 1},
+ {0x0700, 0x0702, 1},
+ {0x07f9, 0x07f9, 1},
+ {0x0964, 0x0965, 1},
+ {0x104a, 0x104b, 1},
+ {0x1362, 0x1362, 1},
+ {0x1367, 0x1368, 1},
+ {0x166e, 0x166e, 1},
+ {0x1735, 0x1736, 1},
+ {0x1803, 0x1803, 1},
+ {0x1809, 0x1809, 1},
+ {0x1944, 0x1945, 1},
+ {0x1aa8, 0x1aab, 1},
+ {0x1b5a, 0x1b5b, 1},
+ {0x1b5e, 0x1b5f, 1},
+ {0x1c3b, 0x1c3c, 1},
+ {0x1c7e, 0x1c7f, 1},
+ {0x203c, 0x203d, 1},
+ {0x2047, 0x2049, 1},
+ {0x2e2e, 0x2e2e, 1},
+ {0x2e3c, 0x2e3c, 1},
+ {0x3002, 0x3002, 1},
+ {0xa4ff, 0xa4ff, 1},
+ {0xa60e, 0xa60f, 1},
+ {0xa6f3, 0xa6f3, 1},
+ {0xa6f7, 0xa6f7, 1},
+ {0xa876, 0xa877, 1},
+ {0xa8ce, 0xa8cf, 1},
+ {0xa92f, 0xa92f, 1},
+ {0xa9c8, 0xa9c9, 1},
+ {0xaa5d, 0xaa5f, 1},
+ {0xaaf0, 0xaaf1, 1},
+ {0xabeb, 0xabeb, 1},
+ {0xfe52, 0xfe52, 1},
+ {0xfe56, 0xfe57, 1},
+ {0xff01, 0xff01, 1},
+ {0xff0e, 0xff0e, 1},
+ {0xff1f, 0xff1f, 1},
+ {0xff61, 0xff61, 1},
+ },
+ R32: []Range32{
+ {0x10a56, 0x10a57, 1},
+ {0x11047, 0x11048, 1},
+ {0x110be, 0x110c1, 1},
+ {0x11141, 0x11143, 1},
+ {0x111c5, 0x111c6, 1},
+ {0x111cd, 0x111cd, 1},
+ {0x11238, 0x11239, 1},
+ {0x1123b, 0x1123c, 1},
+ {0x115c2, 0x115c3, 1},
+ {0x115c9, 0x115c9, 1},
+ {0x11641, 0x11642, 1},
+ {0x16a6e, 0x16a6f, 1},
+ {0x16af5, 0x16af5, 1},
+ {0x16b37, 0x16b38, 1},
+ {0x16b44, 0x16b44, 1},
+ {0x1bc9f, 0x1bc9f, 1},
+ },
+ LatinOffset: 3,
+}
+
+var _Soft_Dotted = &RangeTable{
+ R16: []Range16{
+ {0x0069, 0x006a, 1},
+ {0x012f, 0x012f, 1},
+ {0x0249, 0x0249, 1},
+ {0x0268, 0x0268, 1},
+ {0x029d, 0x029d, 1},
+ {0x02b2, 0x02b2, 1},
+ {0x03f3, 0x03f3, 1},
+ {0x0456, 0x0456, 1},
+ {0x0458, 0x0458, 1},
+ {0x1d62, 0x1d62, 1},
+ {0x1d96, 0x1d96, 1},
+ {0x1da4, 0x1da4, 1},
+ {0x1da8, 0x1da8, 1},
+ {0x1e2d, 0x1e2d, 1},
+ {0x1ecb, 0x1ecb, 1},
+ {0x2071, 0x2071, 1},
+ {0x2148, 0x2149, 1},
+ {0x2c7c, 0x2c7c, 1},
+ },
+ R32: []Range32{
+ {0x1d422, 0x1d423, 1},
+ {0x1d456, 0x1d457, 1},
+ {0x1d48a, 0x1d48b, 1},
+ {0x1d4be, 0x1d4bf, 1},
+ {0x1d4f2, 0x1d4f3, 1},
+ {0x1d526, 0x1d527, 1},
+ {0x1d55a, 0x1d55b, 1},
+ {0x1d58e, 0x1d58f, 1},
+ {0x1d5c2, 0x1d5c3, 1},
+ {0x1d5f6, 0x1d5f7, 1},
+ {0x1d62a, 0x1d62b, 1},
+ {0x1d65e, 0x1d65f, 1},
+ {0x1d692, 0x1d693, 1},
+ },
+ LatinOffset: 1,
+}
+
+var _Terminal_Punctuation = &RangeTable{
+ R16: []Range16{
+ {0x0021, 0x0021, 1},
+ {0x002c, 0x002c, 1},
+ {0x002e, 0x002e, 1},
+ {0x003a, 0x003b, 1},
+ {0x003f, 0x003f, 1},
+ {0x037e, 0x037e, 1},
+ {0x0387, 0x0387, 1},
+ {0x0589, 0x0589, 1},
+ {0x05c3, 0x05c3, 1},
+ {0x060c, 0x060c, 1},
+ {0x061b, 0x061b, 1},
+ {0x061f, 0x061f, 1},
+ {0x06d4, 0x06d4, 1},
+ {0x0700, 0x070a, 1},
+ {0x070c, 0x070c, 1},
+ {0x07f8, 0x07f9, 1},
+ {0x0830, 0x083e, 1},
+ {0x085e, 0x085e, 1},
+ {0x0964, 0x0965, 1},
+ {0x0e5a, 0x0e5b, 1},
+ {0x0f08, 0x0f08, 1},
+ {0x0f0d, 0x0f12, 1},
+ {0x104a, 0x104b, 1},
+ {0x1361, 0x1368, 1},
+ {0x166d, 0x166e, 1},
+ {0x16eb, 0x16ed, 1},
+ {0x1735, 0x1736, 1},
+ {0x17d4, 0x17d6, 1},
+ {0x17da, 0x17da, 1},
+ {0x1802, 0x1805, 1},
+ {0x1808, 0x1809, 1},
+ {0x1944, 0x1945, 1},
+ {0x1aa8, 0x1aab, 1},
+ {0x1b5a, 0x1b5b, 1},
+ {0x1b5d, 0x1b5f, 1},
+ {0x1c3b, 0x1c3f, 1},
+ {0x1c7e, 0x1c7f, 1},
+ {0x203c, 0x203d, 1},
+ {0x2047, 0x2049, 1},
+ {0x2e2e, 0x2e2e, 1},
+ {0x2e3c, 0x2e3c, 1},
+ {0x2e41, 0x2e41, 1},
+ {0x3001, 0x3002, 1},
+ {0xa4fe, 0xa4ff, 1},
+ {0xa60d, 0xa60f, 1},
+ {0xa6f3, 0xa6f7, 1},
+ {0xa876, 0xa877, 1},
+ {0xa8ce, 0xa8cf, 1},
+ {0xa92f, 0xa92f, 1},
+ {0xa9c7, 0xa9c9, 1},
+ {0xaa5d, 0xaa5f, 1},
+ {0xaadf, 0xaadf, 1},
+ {0xaaf0, 0xaaf1, 1},
+ {0xabeb, 0xabeb, 1},
+ {0xfe50, 0xfe52, 1},
+ {0xfe54, 0xfe57, 1},
+ {0xff01, 0xff01, 1},
+ {0xff0c, 0xff0c, 1},
+ {0xff0e, 0xff0e, 1},
+ {0xff1a, 0xff1b, 1},
+ {0xff1f, 0xff1f, 1},
+ {0xff61, 0xff61, 1},
+ {0xff64, 0xff64, 1},
+ },
+ R32: []Range32{
+ {0x1039f, 0x1039f, 1},
+ {0x103d0, 0x103d0, 1},
+ {0x10857, 0x10857, 1},
+ {0x1091f, 0x1091f, 1},
+ {0x10a56, 0x10a57, 1},
+ {0x10af0, 0x10af5, 1},
+ {0x10b3a, 0x10b3f, 1},
+ {0x10b99, 0x10b9c, 1},
+ {0x11047, 0x1104d, 1},
+ {0x110be, 0x110c1, 1},
+ {0x11141, 0x11143, 1},
+ {0x111c5, 0x111c6, 1},
+ {0x111cd, 0x111cd, 1},
+ {0x11238, 0x1123c, 1},
+ {0x115c2, 0x115c5, 1},
+ {0x115c9, 0x115c9, 1},
+ {0x11641, 0x11642, 1},
+ {0x12470, 0x12474, 1},
+ {0x16a6e, 0x16a6f, 1},
+ {0x16af5, 0x16af5, 1},
+ {0x16b37, 0x16b39, 1},
+ {0x16b44, 0x16b44, 1},
+ {0x1bc9f, 0x1bc9f, 1},
+ },
+ LatinOffset: 5,
+}
+
+var _Unified_Ideograph = &RangeTable{
+ R16: []Range16{
+ {0x3400, 0x4db5, 1},
+ {0x4e00, 0x9fcc, 1},
+ {0xfa0e, 0xfa0f, 1},
+ {0xfa11, 0xfa11, 1},
+ {0xfa13, 0xfa14, 1},
+ {0xfa1f, 0xfa1f, 1},
+ {0xfa21, 0xfa21, 1},
+ {0xfa23, 0xfa24, 1},
+ {0xfa27, 0xfa29, 1},
+ },
+ R32: []Range32{
+ {0x20000, 0x2a6d6, 1},
+ {0x2a700, 0x2b734, 1},
+ {0x2b740, 0x2b81d, 1},
+ },
+}
+
+var _Variation_Selector = &RangeTable{
+ R16: []Range16{
+ {0x180b, 0x180d, 1},
+ {0xfe00, 0xfe0f, 1},
+ },
+ R32: []Range32{
+ {0xe0100, 0xe01ef, 1},
+ },
+}
+
+var _White_Space = &RangeTable{
+ R16: []Range16{
+ {0x0009, 0x000d, 1},
+ {0x0020, 0x0020, 1},
+ {0x0085, 0x0085, 1},
+ {0x00a0, 0x00a0, 1},
+ {0x1680, 0x1680, 1},
+ {0x2000, 0x200a, 1},
+ {0x2028, 0x2029, 1},
+ {0x202f, 0x202f, 1},
+ {0x205f, 0x205f, 1},
+ {0x3000, 0x3000, 1},
+ },
+ LatinOffset: 4,
+}
+
+// These variables have type *RangeTable.
+var (
+ ASCII_Hex_Digit = _ASCII_Hex_Digit // ASCII_Hex_Digit is the set of Unicode characters with property ASCII_Hex_Digit.
+ Bidi_Control = _Bidi_Control // Bidi_Control is the set of Unicode characters with property Bidi_Control.
+ Dash = _Dash // Dash is the set of Unicode characters with property Dash.
+ Deprecated = _Deprecated // Deprecated is the set of Unicode characters with property Deprecated.
+ Diacritic = _Diacritic // Diacritic is the set of Unicode characters with property Diacritic.
+ Extender = _Extender // Extender is the set of Unicode characters with property Extender.
+ Hex_Digit = _Hex_Digit // Hex_Digit is the set of Unicode characters with property Hex_Digit.
+ Hyphen = _Hyphen // Hyphen is the set of Unicode characters with property Hyphen.
+ IDS_Binary_Operator = _IDS_Binary_Operator // IDS_Binary_Operator is the set of Unicode characters with property IDS_Binary_Operator.
+ IDS_Trinary_Operator = _IDS_Trinary_Operator // IDS_Trinary_Operator is the set of Unicode characters with property IDS_Trinary_Operator.
+ Ideographic = _Ideographic // Ideographic is the set of Unicode characters with property Ideographic.
+ Join_Control = _Join_Control // Join_Control is the set of Unicode characters with property Join_Control.
+ Logical_Order_Exception = _Logical_Order_Exception // Logical_Order_Exception is the set of Unicode characters with property Logical_Order_Exception.
+ Noncharacter_Code_Point = _Noncharacter_Code_Point // Noncharacter_Code_Point is the set of Unicode characters with property Noncharacter_Code_Point.
+ Other_Alphabetic = _Other_Alphabetic // Other_Alphabetic is the set of Unicode characters with property Other_Alphabetic.
+ Other_Default_Ignorable_Code_Point = _Other_Default_Ignorable_Code_Point // Other_Default_Ignorable_Code_Point is the set of Unicode characters with property Other_Default_Ignorable_Code_Point.
+ Other_Grapheme_Extend = _Other_Grapheme_Extend // Other_Grapheme_Extend is the set of Unicode characters with property Other_Grapheme_Extend.
+ Other_ID_Continue = _Other_ID_Continue // Other_ID_Continue is the set of Unicode characters with property Other_ID_Continue.
+ Other_ID_Start = _Other_ID_Start // Other_ID_Start is the set of Unicode characters with property Other_ID_Start.
+ Other_Lowercase = _Other_Lowercase // Other_Lowercase is the set of Unicode characters with property Other_Lowercase.
+ Other_Math = _Other_Math // Other_Math is the set of Unicode characters with property Other_Math.
+ Other_Uppercase = _Other_Uppercase // Other_Uppercase is the set of Unicode characters with property Other_Uppercase.
+ Pattern_Syntax = _Pattern_Syntax // Pattern_Syntax is the set of Unicode characters with property Pattern_Syntax.
+ Pattern_White_Space = _Pattern_White_Space // Pattern_White_Space is the set of Unicode characters with property Pattern_White_Space.
+ Quotation_Mark = _Quotation_Mark // Quotation_Mark is the set of Unicode characters with property Quotation_Mark.
+ Radical = _Radical // Radical is the set of Unicode characters with property Radical.
+ STerm = _STerm // STerm is the set of Unicode characters with property STerm.
+ Soft_Dotted = _Soft_Dotted // Soft_Dotted is the set of Unicode characters with property Soft_Dotted.
+ Terminal_Punctuation = _Terminal_Punctuation // Terminal_Punctuation is the set of Unicode characters with property Terminal_Punctuation.
+ Unified_Ideograph = _Unified_Ideograph // Unified_Ideograph is the set of Unicode characters with property Unified_Ideograph.
+ Variation_Selector = _Variation_Selector // Variation_Selector is the set of Unicode characters with property Variation_Selector.
+ White_Space = _White_Space // White_Space is the set of Unicode characters with property White_Space.
+)
+
+// Generated by running
+// maketables --data=http://www.unicode.org/Public/7.0.0/ucd/UnicodeData.txt --casefolding=http://www.unicode.org/Public/7.0.0/ucd/CaseFolding.txt
+// DO NOT EDIT
+
+// CaseRanges is the table describing case mappings for all letters with
+// non-self mappings.
+var CaseRanges = _CaseRanges
+var _CaseRanges = []CaseRange{
+ {0x0041, 0x005A, d{0, 32, 0}},
+ {0x0061, 0x007A, d{-32, 0, -32}},
+ {0x00B5, 0x00B5, d{743, 0, 743}},
+ {0x00C0, 0x00D6, d{0, 32, 0}},
+ {0x00D8, 0x00DE, d{0, 32, 0}},
+ {0x00E0, 0x00F6, d{-32, 0, -32}},
+ {0x00F8, 0x00FE, d{-32, 0, -32}},
+ {0x00FF, 0x00FF, d{121, 0, 121}},
+ {0x0100, 0x012F, d{UpperLower, UpperLower, UpperLower}},
+ {0x0130, 0x0130, d{0, -199, 0}},
+ {0x0131, 0x0131, d{-232, 0, -232}},
+ {0x0132, 0x0137, d{UpperLower, UpperLower, UpperLower}},
+ {0x0139, 0x0148, d{UpperLower, UpperLower, UpperLower}},
+ {0x014A, 0x0177, d{UpperLower, UpperLower, UpperLower}},
+ {0x0178, 0x0178, d{0, -121, 0}},
+ {0x0179, 0x017E, d{UpperLower, UpperLower, UpperLower}},
+ {0x017F, 0x017F, d{-300, 0, -300}},
+ {0x0180, 0x0180, d{195, 0, 195}},
+ {0x0181, 0x0181, d{0, 210, 0}},
+ {0x0182, 0x0185, d{UpperLower, UpperLower, UpperLower}},
+ {0x0186, 0x0186, d{0, 206, 0}},
+ {0x0187, 0x0188, d{UpperLower, UpperLower, UpperLower}},
+ {0x0189, 0x018A, d{0, 205, 0}},
+ {0x018B, 0x018C, d{UpperLower, UpperLower, UpperLower}},
+ {0x018E, 0x018E, d{0, 79, 0}},
+ {0x018F, 0x018F, d{0, 202, 0}},
+ {0x0190, 0x0190, d{0, 203, 0}},
+ {0x0191, 0x0192, d{UpperLower, UpperLower, UpperLower}},
+ {0x0193, 0x0193, d{0, 205, 0}},
+ {0x0194, 0x0194, d{0, 207, 0}},
+ {0x0195, 0x0195, d{97, 0, 97}},
+ {0x0196, 0x0196, d{0, 211, 0}},
+ {0x0197, 0x0197, d{0, 209, 0}},
+ {0x0198, 0x0199, d{UpperLower, UpperLower, UpperLower}},
+ {0x019A, 0x019A, d{163, 0, 163}},
+ {0x019C, 0x019C, d{0, 211, 0}},
+ {0x019D, 0x019D, d{0, 213, 0}},
+ {0x019E, 0x019E, d{130, 0, 130}},
+ {0x019F, 0x019F, d{0, 214, 0}},
+ {0x01A0, 0x01A5, d{UpperLower, UpperLower, UpperLower}},
+ {0x01A6, 0x01A6, d{0, 218, 0}},
+ {0x01A7, 0x01A8, d{UpperLower, UpperLower, UpperLower}},
+ {0x01A9, 0x01A9, d{0, 218, 0}},
+ {0x01AC, 0x01AD, d{UpperLower, UpperLower, UpperLower}},
+ {0x01AE, 0x01AE, d{0, 218, 0}},
+ {0x01AF, 0x01B0, d{UpperLower, UpperLower, UpperLower}},
+ {0x01B1, 0x01B2, d{0, 217, 0}},
+ {0x01B3, 0x01B6, d{UpperLower, UpperLower, UpperLower}},
+ {0x01B7, 0x01B7, d{0, 219, 0}},
+ {0x01B8, 0x01B9, d{UpperLower, UpperLower, UpperLower}},
+ {0x01BC, 0x01BD, d{UpperLower, UpperLower, UpperLower}},
+ {0x01BF, 0x01BF, d{56, 0, 56}},
+ {0x01C4, 0x01C4, d{0, 2, 1}},
+ {0x01C5, 0x01C5, d{-1, 1, 0}},
+ {0x01C6, 0x01C6, d{-2, 0, -1}},
+ {0x01C7, 0x01C7, d{0, 2, 1}},
+ {0x01C8, 0x01C8, d{-1, 1, 0}},
+ {0x01C9, 0x01C9, d{-2, 0, -1}},
+ {0x01CA, 0x01CA, d{0, 2, 1}},
+ {0x01CB, 0x01CB, d{-1, 1, 0}},
+ {0x01CC, 0x01CC, d{-2, 0, -1}},
+ {0x01CD, 0x01DC, d{UpperLower, UpperLower, UpperLower}},
+ {0x01DD, 0x01DD, d{-79, 0, -79}},
+ {0x01DE, 0x01EF, d{UpperLower, UpperLower, UpperLower}},
+ {0x01F1, 0x01F1, d{0, 2, 1}},
+ {0x01F2, 0x01F2, d{-1, 1, 0}},
+ {0x01F3, 0x01F3, d{-2, 0, -1}},
+ {0x01F4, 0x01F5, d{UpperLower, UpperLower, UpperLower}},
+ {0x01F6, 0x01F6, d{0, -97, 0}},
+ {0x01F7, 0x01F7, d{0, -56, 0}},
+ {0x01F8, 0x021F, d{UpperLower, UpperLower, UpperLower}},
+ {0x0220, 0x0220, d{0, -130, 0}},
+ {0x0222, 0x0233, d{UpperLower, UpperLower, UpperLower}},
+ {0x023A, 0x023A, d{0, 10795, 0}},
+ {0x023B, 0x023C, d{UpperLower, UpperLower, UpperLower}},
+ {0x023D, 0x023D, d{0, -163, 0}},
+ {0x023E, 0x023E, d{0, 10792, 0}},
+ {0x023F, 0x0240, d{10815, 0, 10815}},
+ {0x0241, 0x0242, d{UpperLower, UpperLower, UpperLower}},
+ {0x0243, 0x0243, d{0, -195, 0}},
+ {0x0244, 0x0244, d{0, 69, 0}},
+ {0x0245, 0x0245, d{0, 71, 0}},
+ {0x0246, 0x024F, d{UpperLower, UpperLower, UpperLower}},
+ {0x0250, 0x0250, d{10783, 0, 10783}},
+ {0x0251, 0x0251, d{10780, 0, 10780}},
+ {0x0252, 0x0252, d{10782, 0, 10782}},
+ {0x0253, 0x0253, d{-210, 0, -210}},
+ {0x0254, 0x0254, d{-206, 0, -206}},
+ {0x0256, 0x0257, d{-205, 0, -205}},
+ {0x0259, 0x0259, d{-202, 0, -202}},
+ {0x025B, 0x025B, d{-203, 0, -203}},
+ {0x025C, 0x025C, d{42319, 0, 42319}},
+ {0x0260, 0x0260, d{-205, 0, -205}},
+ {0x0261, 0x0261, d{42315, 0, 42315}},
+ {0x0263, 0x0263, d{-207, 0, -207}},
+ {0x0265, 0x0265, d{42280, 0, 42280}},
+ {0x0266, 0x0266, d{42308, 0, 42308}},
+ {0x0268, 0x0268, d{-209, 0, -209}},
+ {0x0269, 0x0269, d{-211, 0, -211}},
+ {0x026B, 0x026B, d{10743, 0, 10743}},
+ {0x026C, 0x026C, d{42305, 0, 42305}},
+ {0x026F, 0x026F, d{-211, 0, -211}},
+ {0x0271, 0x0271, d{10749, 0, 10749}},
+ {0x0272, 0x0272, d{-213, 0, -213}},
+ {0x0275, 0x0275, d{-214, 0, -214}},
+ {0x027D, 0x027D, d{10727, 0, 10727}},
+ {0x0280, 0x0280, d{-218, 0, -218}},
+ {0x0283, 0x0283, d{-218, 0, -218}},
+ {0x0287, 0x0287, d{42282, 0, 42282}},
+ {0x0288, 0x0288, d{-218, 0, -218}},
+ {0x0289, 0x0289, d{-69, 0, -69}},
+ {0x028A, 0x028B, d{-217, 0, -217}},
+ {0x028C, 0x028C, d{-71, 0, -71}},
+ {0x0292, 0x0292, d{-219, 0, -219}},
+ {0x029E, 0x029E, d{42258, 0, 42258}},
+ {0x0345, 0x0345, d{84, 0, 84}},
+ {0x0370, 0x0373, d{UpperLower, UpperLower, UpperLower}},
+ {0x0376, 0x0377, d{UpperLower, UpperLower, UpperLower}},
+ {0x037B, 0x037D, d{130, 0, 130}},
+ {0x037F, 0x037F, d{0, 116, 0}},
+ {0x0386, 0x0386, d{0, 38, 0}},
+ {0x0388, 0x038A, d{0, 37, 0}},
+ {0x038C, 0x038C, d{0, 64, 0}},
+ {0x038E, 0x038F, d{0, 63, 0}},
+ {0x0391, 0x03A1, d{0, 32, 0}},
+ {0x03A3, 0x03AB, d{0, 32, 0}},
+ {0x03AC, 0x03AC, d{-38, 0, -38}},
+ {0x03AD, 0x03AF, d{-37, 0, -37}},
+ {0x03B1, 0x03C1, d{-32, 0, -32}},
+ {0x03C2, 0x03C2, d{-31, 0, -31}},
+ {0x03C3, 0x03CB, d{-32, 0, -32}},
+ {0x03CC, 0x03CC, d{-64, 0, -64}},
+ {0x03CD, 0x03CE, d{-63, 0, -63}},
+ {0x03CF, 0x03CF, d{0, 8, 0}},
+ {0x03D0, 0x03D0, d{-62, 0, -62}},
+ {0x03D1, 0x03D1, d{-57, 0, -57}},
+ {0x03D5, 0x03D5, d{-47, 0, -47}},
+ {0x03D6, 0x03D6, d{-54, 0, -54}},
+ {0x03D7, 0x03D7, d{-8, 0, -8}},
+ {0x03D8, 0x03EF, d{UpperLower, UpperLower, UpperLower}},
+ {0x03F0, 0x03F0, d{-86, 0, -86}},
+ {0x03F1, 0x03F1, d{-80, 0, -80}},
+ {0x03F2, 0x03F2, d{7, 0, 7}},
+ {0x03F3, 0x03F3, d{-116, 0, -116}},
+ {0x03F4, 0x03F4, d{0, -60, 0}},
+ {0x03F5, 0x03F5, d{-96, 0, -96}},
+ {0x03F7, 0x03F8, d{UpperLower, UpperLower, UpperLower}},
+ {0x03F9, 0x03F9, d{0, -7, 0}},
+ {0x03FA, 0x03FB, d{UpperLower, UpperLower, UpperLower}},
+ {0x03FD, 0x03FF, d{0, -130, 0}},
+ {0x0400, 0x040F, d{0, 80, 0}},
+ {0x0410, 0x042F, d{0, 32, 0}},
+ {0x0430, 0x044F, d{-32, 0, -32}},
+ {0x0450, 0x045F, d{-80, 0, -80}},
+ {0x0460, 0x0481, d{UpperLower, UpperLower, UpperLower}},
+ {0x048A, 0x04BF, d{UpperLower, UpperLower, UpperLower}},
+ {0x04C0, 0x04C0, d{0, 15, 0}},
+ {0x04C1, 0x04CE, d{UpperLower, UpperLower, UpperLower}},
+ {0x04CF, 0x04CF, d{-15, 0, -15}},
+ {0x04D0, 0x052F, d{UpperLower, UpperLower, UpperLower}},
+ {0x0531, 0x0556, d{0, 48, 0}},
+ {0x0561, 0x0586, d{-48, 0, -48}},
+ {0x10A0, 0x10C5, d{0, 7264, 0}},
+ {0x10C7, 0x10C7, d{0, 7264, 0}},
+ {0x10CD, 0x10CD, d{0, 7264, 0}},
+ {0x1D79, 0x1D79, d{35332, 0, 35332}},
+ {0x1D7D, 0x1D7D, d{3814, 0, 3814}},
+ {0x1E00, 0x1E95, d{UpperLower, UpperLower, UpperLower}},
+ {0x1E9B, 0x1E9B, d{-59, 0, -59}},
+ {0x1E9E, 0x1E9E, d{0, -7615, 0}},
+ {0x1EA0, 0x1EFF, d{UpperLower, UpperLower, UpperLower}},
+ {0x1F00, 0x1F07, d{8, 0, 8}},
+ {0x1F08, 0x1F0F, d{0, -8, 0}},
+ {0x1F10, 0x1F15, d{8, 0, 8}},
+ {0x1F18, 0x1F1D, d{0, -8, 0}},
+ {0x1F20, 0x1F27, d{8, 0, 8}},
+ {0x1F28, 0x1F2F, d{0, -8, 0}},
+ {0x1F30, 0x1F37, d{8, 0, 8}},
+ {0x1F38, 0x1F3F, d{0, -8, 0}},
+ {0x1F40, 0x1F45, d{8, 0, 8}},
+ {0x1F48, 0x1F4D, d{0, -8, 0}},
+ {0x1F51, 0x1F51, d{8, 0, 8}},
+ {0x1F53, 0x1F53, d{8, 0, 8}},
+ {0x1F55, 0x1F55, d{8, 0, 8}},
+ {0x1F57, 0x1F57, d{8, 0, 8}},
+ {0x1F59, 0x1F59, d{0, -8, 0}},
+ {0x1F5B, 0x1F5B, d{0, -8, 0}},
+ {0x1F5D, 0x1F5D, d{0, -8, 0}},
+ {0x1F5F, 0x1F5F, d{0, -8, 0}},
+ {0x1F60, 0x1F67, d{8, 0, 8}},
+ {0x1F68, 0x1F6F, d{0, -8, 0}},
+ {0x1F70, 0x1F71, d{74, 0, 74}},
+ {0x1F72, 0x1F75, d{86, 0, 86}},
+ {0x1F76, 0x1F77, d{100, 0, 100}},
+ {0x1F78, 0x1F79, d{128, 0, 128}},
+ {0x1F7A, 0x1F7B, d{112, 0, 112}},
+ {0x1F7C, 0x1F7D, d{126, 0, 126}},
+ {0x1F80, 0x1F87, d{8, 0, 8}},
+ {0x1F88, 0x1F8F, d{0, -8, 0}},
+ {0x1F90, 0x1F97, d{8, 0, 8}},
+ {0x1F98, 0x1F9F, d{0, -8, 0}},
+ {0x1FA0, 0x1FA7, d{8, 0, 8}},
+ {0x1FA8, 0x1FAF, d{0, -8, 0}},
+ {0x1FB0, 0x1FB1, d{8, 0, 8}},
+ {0x1FB3, 0x1FB3, d{9, 0, 9}},
+ {0x1FB8, 0x1FB9, d{0, -8, 0}},
+ {0x1FBA, 0x1FBB, d{0, -74, 0}},
+ {0x1FBC, 0x1FBC, d{0, -9, 0}},
+ {0x1FBE, 0x1FBE, d{-7205, 0, -7205}},
+ {0x1FC3, 0x1FC3, d{9, 0, 9}},
+ {0x1FC8, 0x1FCB, d{0, -86, 0}},
+ {0x1FCC, 0x1FCC, d{0, -9, 0}},
+ {0x1FD0, 0x1FD1, d{8, 0, 8}},
+ {0x1FD8, 0x1FD9, d{0, -8, 0}},
+ {0x1FDA, 0x1FDB, d{0, -100, 0}},
+ {0x1FE0, 0x1FE1, d{8, 0, 8}},
+ {0x1FE5, 0x1FE5, d{7, 0, 7}},
+ {0x1FE8, 0x1FE9, d{0, -8, 0}},
+ {0x1FEA, 0x1FEB, d{0, -112, 0}},
+ {0x1FEC, 0x1FEC, d{0, -7, 0}},
+ {0x1FF3, 0x1FF3, d{9, 0, 9}},
+ {0x1FF8, 0x1FF9, d{0, -128, 0}},
+ {0x1FFA, 0x1FFB, d{0, -126, 0}},
+ {0x1FFC, 0x1FFC, d{0, -9, 0}},
+ {0x2126, 0x2126, d{0, -7517, 0}},
+ {0x212A, 0x212A, d{0, -8383, 0}},
+ {0x212B, 0x212B, d{0, -8262, 0}},
+ {0x2132, 0x2132, d{0, 28, 0}},
+ {0x214E, 0x214E, d{-28, 0, -28}},
+ {0x2160, 0x216F, d{0, 16, 0}},
+ {0x2170, 0x217F, d{-16, 0, -16}},
+ {0x2183, 0x2184, d{UpperLower, UpperLower, UpperLower}},
+ {0x24B6, 0x24CF, d{0, 26, 0}},
+ {0x24D0, 0x24E9, d{-26, 0, -26}},
+ {0x2C00, 0x2C2E, d{0, 48, 0}},
+ {0x2C30, 0x2C5E, d{-48, 0, -48}},
+ {0x2C60, 0x2C61, d{UpperLower, UpperLower, UpperLower}},
+ {0x2C62, 0x2C62, d{0, -10743, 0}},
+ {0x2C63, 0x2C63, d{0, -3814, 0}},
+ {0x2C64, 0x2C64, d{0, -10727, 0}},
+ {0x2C65, 0x2C65, d{-10795, 0, -10795}},
+ {0x2C66, 0x2C66, d{-10792, 0, -10792}},
+ {0x2C67, 0x2C6C, d{UpperLower, UpperLower, UpperLower}},
+ {0x2C6D, 0x2C6D, d{0, -10780, 0}},
+ {0x2C6E, 0x2C6E, d{0, -10749, 0}},
+ {0x2C6F, 0x2C6F, d{0, -10783, 0}},
+ {0x2C70, 0x2C70, d{0, -10782, 0}},
+ {0x2C72, 0x2C73, d{UpperLower, UpperLower, UpperLower}},
+ {0x2C75, 0x2C76, d{UpperLower, UpperLower, UpperLower}},
+ {0x2C7E, 0x2C7F, d{0, -10815, 0}},
+ {0x2C80, 0x2CE3, d{UpperLower, UpperLower, UpperLower}},
+ {0x2CEB, 0x2CEE, d{UpperLower, UpperLower, UpperLower}},
+ {0x2CF2, 0x2CF3, d{UpperLower, UpperLower, UpperLower}},
+ {0x2D00, 0x2D25, d{-7264, 0, -7264}},
+ {0x2D27, 0x2D27, d{-7264, 0, -7264}},
+ {0x2D2D, 0x2D2D, d{-7264, 0, -7264}},
+ {0xA640, 0xA66D, d{UpperLower, UpperLower, UpperLower}},
+ {0xA680, 0xA69B, d{UpperLower, UpperLower, UpperLower}},
+ {0xA722, 0xA72F, d{UpperLower, UpperLower, UpperLower}},
+ {0xA732, 0xA76F, d{UpperLower, UpperLower, UpperLower}},
+ {0xA779, 0xA77C, d{UpperLower, UpperLower, UpperLower}},
+ {0xA77D, 0xA77D, d{0, -35332, 0}},
+ {0xA77E, 0xA787, d{UpperLower, UpperLower, UpperLower}},
+ {0xA78B, 0xA78C, d{UpperLower, UpperLower, UpperLower}},
+ {0xA78D, 0xA78D, d{0, -42280, 0}},
+ {0xA790, 0xA793, d{UpperLower, UpperLower, UpperLower}},
+ {0xA796, 0xA7A9, d{UpperLower, UpperLower, UpperLower}},
+ {0xA7AA, 0xA7AA, d{0, -42308, 0}},
+ {0xA7AB, 0xA7AB, d{0, -42319, 0}},
+ {0xA7AC, 0xA7AC, d{0, -42315, 0}},
+ {0xA7AD, 0xA7AD, d{0, -42305, 0}},
+ {0xA7B0, 0xA7B0, d{0, -42258, 0}},
+ {0xA7B1, 0xA7B1, d{0, -42282, 0}},
+ {0xFF21, 0xFF3A, d{0, 32, 0}},
+ {0xFF41, 0xFF5A, d{-32, 0, -32}},
+ {0x10400, 0x10427, d{0, 40, 0}},
+ {0x10428, 0x1044F, d{-40, 0, -40}},
+ {0x118A0, 0x118BF, d{0, 32, 0}},
+ {0x118C0, 0x118DF, d{-32, 0, -32}},
+}
+var properties = [MaxLatin1 + 1]uint8{
+ 0x00: pC, // '\x00'
+ 0x01: pC, // '\x01'
+ 0x02: pC, // '\x02'
+ 0x03: pC, // '\x03'
+ 0x04: pC, // '\x04'
+ 0x05: pC, // '\x05'
+ 0x06: pC, // '\x06'
+ 0x07: pC, // '\a'
+ 0x08: pC, // '\b'
+ 0x09: pC, // '\t'
+ 0x0A: pC, // '\n'
+ 0x0B: pC, // '\v'
+ 0x0C: pC, // '\f'
+ 0x0D: pC, // '\r'
+ 0x0E: pC, // '\x0e'
+ 0x0F: pC, // '\x0f'
+ 0x10: pC, // '\x10'
+ 0x11: pC, // '\x11'
+ 0x12: pC, // '\x12'
+ 0x13: pC, // '\x13'
+ 0x14: pC, // '\x14'
+ 0x15: pC, // '\x15'
+ 0x16: pC, // '\x16'
+ 0x17: pC, // '\x17'
+ 0x18: pC, // '\x18'
+ 0x19: pC, // '\x19'
+ 0x1A: pC, // '\x1a'
+ 0x1B: pC, // '\x1b'
+ 0x1C: pC, // '\x1c'
+ 0x1D: pC, // '\x1d'
+ 0x1E: pC, // '\x1e'
+ 0x1F: pC, // '\x1f'
+ 0x20: pZ | pp, // ' '
+ 0x21: pP | pp, // '!'
+ 0x22: pP | pp, // '"'
+ 0x23: pP | pp, // '#'
+ 0x24: pS | pp, // '$'
+ 0x25: pP | pp, // '%'
+ 0x26: pP | pp, // '&'
+ 0x27: pP | pp, // '\''
+ 0x28: pP | pp, // '('
+ 0x29: pP | pp, // ')'
+ 0x2A: pP | pp, // '*'
+ 0x2B: pS | pp, // '+'
+ 0x2C: pP | pp, // ','
+ 0x2D: pP | pp, // '-'
+ 0x2E: pP | pp, // '.'
+ 0x2F: pP | pp, // '/'
+ 0x30: pN | pp, // '0'
+ 0x31: pN | pp, // '1'
+ 0x32: pN | pp, // '2'
+ 0x33: pN | pp, // '3'
+ 0x34: pN | pp, // '4'
+ 0x35: pN | pp, // '5'
+ 0x36: pN | pp, // '6'
+ 0x37: pN | pp, // '7'
+ 0x38: pN | pp, // '8'
+ 0x39: pN | pp, // '9'
+ 0x3A: pP | pp, // ':'
+ 0x3B: pP | pp, // ';'
+ 0x3C: pS | pp, // '<'
+ 0x3D: pS | pp, // '='
+ 0x3E: pS | pp, // '>'
+ 0x3F: pP | pp, // '?'
+ 0x40: pP | pp, // '@'
+ 0x41: pLu | pp, // 'A'
+ 0x42: pLu | pp, // 'B'
+ 0x43: pLu | pp, // 'C'
+ 0x44: pLu | pp, // 'D'
+ 0x45: pLu | pp, // 'E'
+ 0x46: pLu | pp, // 'F'
+ 0x47: pLu | pp, // 'G'
+ 0x48: pLu | pp, // 'H'
+ 0x49: pLu | pp, // 'I'
+ 0x4A: pLu | pp, // 'J'
+ 0x4B: pLu | pp, // 'K'
+ 0x4C: pLu | pp, // 'L'
+ 0x4D: pLu | pp, // 'M'
+ 0x4E: pLu | pp, // 'N'
+ 0x4F: pLu | pp, // 'O'
+ 0x50: pLu | pp, // 'P'
+ 0x51: pLu | pp, // 'Q'
+ 0x52: pLu | pp, // 'R'
+ 0x53: pLu | pp, // 'S'
+ 0x54: pLu | pp, // 'T'
+ 0x55: pLu | pp, // 'U'
+ 0x56: pLu | pp, // 'V'
+ 0x57: pLu | pp, // 'W'
+ 0x58: pLu | pp, // 'X'
+ 0x59: pLu | pp, // 'Y'
+ 0x5A: pLu | pp, // 'Z'
+ 0x5B: pP | pp, // '['
+ 0x5C: pP | pp, // '\\'
+ 0x5D: pP | pp, // ']'
+ 0x5E: pS | pp, // '^'
+ 0x5F: pP | pp, // '_'
+ 0x60: pS | pp, // '`'
+ 0x61: pLl | pp, // 'a'
+ 0x62: pLl | pp, // 'b'
+ 0x63: pLl | pp, // 'c'
+ 0x64: pLl | pp, // 'd'
+ 0x65: pLl | pp, // 'e'
+ 0x66: pLl | pp, // 'f'
+ 0x67: pLl | pp, // 'g'
+ 0x68: pLl | pp, // 'h'
+ 0x69: pLl | pp, // 'i'
+ 0x6A: pLl | pp, // 'j'
+ 0x6B: pLl | pp, // 'k'
+ 0x6C: pLl | pp, // 'l'
+ 0x6D: pLl | pp, // 'm'
+ 0x6E: pLl | pp, // 'n'
+ 0x6F: pLl | pp, // 'o'
+ 0x70: pLl | pp, // 'p'
+ 0x71: pLl | pp, // 'q'
+ 0x72: pLl | pp, // 'r'
+ 0x73: pLl | pp, // 's'
+ 0x74: pLl | pp, // 't'
+ 0x75: pLl | pp, // 'u'
+ 0x76: pLl | pp, // 'v'
+ 0x77: pLl | pp, // 'w'
+ 0x78: pLl | pp, // 'x'
+ 0x79: pLl | pp, // 'y'
+ 0x7A: pLl | pp, // 'z'
+ 0x7B: pP | pp, // '{'
+ 0x7C: pS | pp, // '|'
+ 0x7D: pP | pp, // '}'
+ 0x7E: pS | pp, // '~'
+ 0x7F: pC, // '\u007f'
+ 0x80: pC, // '\u0080'
+ 0x81: pC, // '\u0081'
+ 0x82: pC, // '\u0082'
+ 0x83: pC, // '\u0083'
+ 0x84: pC, // '\u0084'
+ 0x85: pC, // '\u0085'
+ 0x86: pC, // '\u0086'
+ 0x87: pC, // '\u0087'
+ 0x88: pC, // '\u0088'
+ 0x89: pC, // '\u0089'
+ 0x8A: pC, // '\u008a'
+ 0x8B: pC, // '\u008b'
+ 0x8C: pC, // '\u008c'
+ 0x8D: pC, // '\u008d'
+ 0x8E: pC, // '\u008e'
+ 0x8F: pC, // '\u008f'
+ 0x90: pC, // '\u0090'
+ 0x91: pC, // '\u0091'
+ 0x92: pC, // '\u0092'
+ 0x93: pC, // '\u0093'
+ 0x94: pC, // '\u0094'
+ 0x95: pC, // '\u0095'
+ 0x96: pC, // '\u0096'
+ 0x97: pC, // '\u0097'
+ 0x98: pC, // '\u0098'
+ 0x99: pC, // '\u0099'
+ 0x9A: pC, // '\u009a'
+ 0x9B: pC, // '\u009b'
+ 0x9C: pC, // '\u009c'
+ 0x9D: pC, // '\u009d'
+ 0x9E: pC, // '\u009e'
+ 0x9F: pC, // '\u009f'
+ 0xA0: pZ, // '\u00a0'
+ 0xA1: pP | pp, // '¡'
+ 0xA2: pS | pp, // '¢'
+ 0xA3: pS | pp, // '£'
+ 0xA4: pS | pp, // '¤'
+ 0xA5: pS | pp, // '¥'
+ 0xA6: pS | pp, // '¦'
+ 0xA7: pP | pp, // '§'
+ 0xA8: pS | pp, // '¨'
+ 0xA9: pS | pp, // '©'
+ 0xAA: pLo | pp, // 'ª'
+ 0xAB: pP | pp, // '«'
+ 0xAC: pS | pp, // '¬'
+ 0xAD: 0, // '\u00ad'
+ 0xAE: pS | pp, // '®'
+ 0xAF: pS | pp, // '¯'
+ 0xB0: pS | pp, // '°'
+ 0xB1: pS | pp, // '±'
+ 0xB2: pN | pp, // '²'
+ 0xB3: pN | pp, // '³'
+ 0xB4: pS | pp, // '´'
+ 0xB5: pLl | pp, // 'µ'
+ 0xB6: pP | pp, // '¶'
+ 0xB7: pP | pp, // '·'
+ 0xB8: pS | pp, // '¸'
+ 0xB9: pN | pp, // '¹'
+ 0xBA: pLo | pp, // 'º'
+ 0xBB: pP | pp, // '»'
+ 0xBC: pN | pp, // '¼'
+ 0xBD: pN | pp, // '½'
+ 0xBE: pN | pp, // '¾'
+ 0xBF: pP | pp, // '¿'
+ 0xC0: pLu | pp, // 'À'
+ 0xC1: pLu | pp, // 'Á'
+ 0xC2: pLu | pp, // 'Â'
+ 0xC3: pLu | pp, // 'Ã'
+ 0xC4: pLu | pp, // 'Ä'
+ 0xC5: pLu | pp, // 'Å'
+ 0xC6: pLu | pp, // 'Æ'
+ 0xC7: pLu | pp, // 'Ç'
+ 0xC8: pLu | pp, // 'È'
+ 0xC9: pLu | pp, // 'É'
+ 0xCA: pLu | pp, // 'Ê'
+ 0xCB: pLu | pp, // 'Ë'
+ 0xCC: pLu | pp, // 'Ì'
+ 0xCD: pLu | pp, // 'Í'
+ 0xCE: pLu | pp, // 'Î'
+ 0xCF: pLu | pp, // 'Ï'
+ 0xD0: pLu | pp, // 'Ð'
+ 0xD1: pLu | pp, // 'Ñ'
+ 0xD2: pLu | pp, // 'Ò'
+ 0xD3: pLu | pp, // 'Ó'
+ 0xD4: pLu | pp, // 'Ô'
+ 0xD5: pLu | pp, // 'Õ'
+ 0xD6: pLu | pp, // 'Ö'
+ 0xD7: pS | pp, // '×'
+ 0xD8: pLu | pp, // 'Ø'
+ 0xD9: pLu | pp, // 'Ù'
+ 0xDA: pLu | pp, // 'Ú'
+ 0xDB: pLu | pp, // 'Û'
+ 0xDC: pLu | pp, // 'Ü'
+ 0xDD: pLu | pp, // 'Ý'
+ 0xDE: pLu | pp, // 'Þ'
+ 0xDF: pLl | pp, // 'ß'
+ 0xE0: pLl | pp, // 'à'
+ 0xE1: pLl | pp, // 'á'
+ 0xE2: pLl | pp, // 'â'
+ 0xE3: pLl | pp, // 'ã'
+ 0xE4: pLl | pp, // 'ä'
+ 0xE5: pLl | pp, // 'å'
+ 0xE6: pLl | pp, // 'æ'
+ 0xE7: pLl | pp, // 'ç'
+ 0xE8: pLl | pp, // 'è'
+ 0xE9: pLl | pp, // 'é'
+ 0xEA: pLl | pp, // 'ê'
+ 0xEB: pLl | pp, // 'ë'
+ 0xEC: pLl | pp, // 'ì'
+ 0xED: pLl | pp, // 'í'
+ 0xEE: pLl | pp, // 'î'
+ 0xEF: pLl | pp, // 'ï'
+ 0xF0: pLl | pp, // 'ð'
+ 0xF1: pLl | pp, // 'ñ'
+ 0xF2: pLl | pp, // 'ò'
+ 0xF3: pLl | pp, // 'ó'
+ 0xF4: pLl | pp, // 'ô'
+ 0xF5: pLl | pp, // 'õ'
+ 0xF6: pLl | pp, // 'ö'
+ 0xF7: pS | pp, // '÷'
+ 0xF8: pLl | pp, // 'ø'
+ 0xF9: pLl | pp, // 'ù'
+ 0xFA: pLl | pp, // 'ú'
+ 0xFB: pLl | pp, // 'û'
+ 0xFC: pLl | pp, // 'ü'
+ 0xFD: pLl | pp, // 'ý'
+ 0xFE: pLl | pp, // 'þ'
+ 0xFF: pLl | pp, // 'ÿ'
+}
+
+var caseOrbit = []foldPair{
+ {0x004B, 0x006B},
+ {0x0053, 0x0073},
+ {0x006B, 0x212A},
+ {0x0073, 0x017F},
+ {0x00B5, 0x039C},
+ {0x00C5, 0x00E5},
+ {0x00DF, 0x1E9E},
+ {0x00E5, 0x212B},
+ {0x0130, 0x0130},
+ {0x0131, 0x0131},
+ {0x017F, 0x0053},
+ {0x01C4, 0x01C5},
+ {0x01C5, 0x01C6},
+ {0x01C6, 0x01C4},
+ {0x01C7, 0x01C8},
+ {0x01C8, 0x01C9},
+ {0x01C9, 0x01C7},
+ {0x01CA, 0x01CB},
+ {0x01CB, 0x01CC},
+ {0x01CC, 0x01CA},
+ {0x01F1, 0x01F2},
+ {0x01F2, 0x01F3},
+ {0x01F3, 0x01F1},
+ {0x0345, 0x0399},
+ {0x0392, 0x03B2},
+ {0x0395, 0x03B5},
+ {0x0398, 0x03B8},
+ {0x0399, 0x03B9},
+ {0x039A, 0x03BA},
+ {0x039C, 0x03BC},
+ {0x03A0, 0x03C0},
+ {0x03A1, 0x03C1},
+ {0x03A3, 0x03C2},
+ {0x03A6, 0x03C6},
+ {0x03A9, 0x03C9},
+ {0x03B2, 0x03D0},
+ {0x03B5, 0x03F5},
+ {0x03B8, 0x03D1},
+ {0x03B9, 0x1FBE},
+ {0x03BA, 0x03F0},
+ {0x03BC, 0x00B5},
+ {0x03C0, 0x03D6},
+ {0x03C1, 0x03F1},
+ {0x03C2, 0x03C3},
+ {0x03C3, 0x03A3},
+ {0x03C6, 0x03D5},
+ {0x03C9, 0x2126},
+ {0x03D0, 0x0392},
+ {0x03D1, 0x03F4},
+ {0x03D5, 0x03A6},
+ {0x03D6, 0x03A0},
+ {0x03F0, 0x039A},
+ {0x03F1, 0x03A1},
+ {0x03F4, 0x0398},
+ {0x03F5, 0x0395},
+ {0x1E60, 0x1E61},
+ {0x1E61, 0x1E9B},
+ {0x1E9B, 0x1E60},
+ {0x1E9E, 0x00DF},
+ {0x1FBE, 0x0345},
+ {0x2126, 0x03A9},
+ {0x212A, 0x004B},
+ {0x212B, 0x00C5},
+}
+
+// FoldCategory maps a category name to a table of
+// code points outside the category that are equivalent under
+// simple case folding to code points inside the category.
+// If there is no entry for a category name, there are no such points.
+var FoldCategory = map[string]*RangeTable{
+ "Common": foldCommon,
+ "Greek": foldGreek,
+ "Inherited": foldInherited,
+ "L": foldL,
+ "Ll": foldLl,
+ "Lt": foldLt,
+ "Lu": foldLu,
+ "M": foldM,
+ "Mn": foldMn,
+}
+
+var foldCommon = &RangeTable{
+ R16: []Range16{
+ {0x039c, 0x03bc, 32},
+ },
+}
+
+var foldGreek = &RangeTable{
+ R16: []Range16{
+ {0x00b5, 0x0345, 656},
+ },
+}
+
+var foldInherited = &RangeTable{
+ R16: []Range16{
+ {0x0399, 0x03b9, 32},
+ {0x1fbe, 0x1fbe, 1},
+ },
+}
+
+var foldL = &RangeTable{
+ R16: []Range16{
+ {0x0345, 0x0345, 1},
+ },
+}
+
+var foldLl = &RangeTable{
+ R16: []Range16{
+ {0x0041, 0x005a, 1},
+ {0x00c0, 0x00d6, 1},
+ {0x00d8, 0x00de, 1},
+ {0x0100, 0x012e, 2},
+ {0x0132, 0x0136, 2},
+ {0x0139, 0x0147, 2},
+ {0x014a, 0x0178, 2},
+ {0x0179, 0x017d, 2},
+ {0x0181, 0x0182, 1},
+ {0x0184, 0x0186, 2},
+ {0x0187, 0x0189, 2},
+ {0x018a, 0x018b, 1},
+ {0x018e, 0x0191, 1},
+ {0x0193, 0x0194, 1},
+ {0x0196, 0x0198, 1},
+ {0x019c, 0x019d, 1},
+ {0x019f, 0x01a0, 1},
+ {0x01a2, 0x01a6, 2},
+ {0x01a7, 0x01a9, 2},
+ {0x01ac, 0x01ae, 2},
+ {0x01af, 0x01b1, 2},
+ {0x01b2, 0x01b3, 1},
+ {0x01b5, 0x01b7, 2},
+ {0x01b8, 0x01bc, 4},
+ {0x01c4, 0x01c5, 1},
+ {0x01c7, 0x01c8, 1},
+ {0x01ca, 0x01cb, 1},
+ {0x01cd, 0x01db, 2},
+ {0x01de, 0x01ee, 2},
+ {0x01f1, 0x01f2, 1},
+ {0x01f4, 0x01f6, 2},
+ {0x01f7, 0x01f8, 1},
+ {0x01fa, 0x0232, 2},
+ {0x023a, 0x023b, 1},
+ {0x023d, 0x023e, 1},
+ {0x0241, 0x0243, 2},
+ {0x0244, 0x0246, 1},
+ {0x0248, 0x024e, 2},
+ {0x0345, 0x0370, 43},
+ {0x0372, 0x0376, 4},
+ {0x037f, 0x0386, 7},
+ {0x0388, 0x038a, 1},
+ {0x038c, 0x038e, 2},
+ {0x038f, 0x0391, 2},
+ {0x0392, 0x03a1, 1},
+ {0x03a3, 0x03ab, 1},
+ {0x03cf, 0x03d8, 9},
+ {0x03da, 0x03ee, 2},
+ {0x03f4, 0x03f7, 3},
+ {0x03f9, 0x03fa, 1},
+ {0x03fd, 0x042f, 1},
+ {0x0460, 0x0480, 2},
+ {0x048a, 0x04c0, 2},
+ {0x04c1, 0x04cd, 2},
+ {0x04d0, 0x052e, 2},
+ {0x0531, 0x0556, 1},
+ {0x10a0, 0x10c5, 1},
+ {0x10c7, 0x10cd, 6},
+ {0x1e00, 0x1e94, 2},
+ {0x1e9e, 0x1efe, 2},
+ {0x1f08, 0x1f0f, 1},
+ {0x1f18, 0x1f1d, 1},
+ {0x1f28, 0x1f2f, 1},
+ {0x1f38, 0x1f3f, 1},
+ {0x1f48, 0x1f4d, 1},
+ {0x1f59, 0x1f5f, 2},
+ {0x1f68, 0x1f6f, 1},
+ {0x1f88, 0x1f8f, 1},
+ {0x1f98, 0x1f9f, 1},
+ {0x1fa8, 0x1faf, 1},
+ {0x1fb8, 0x1fbc, 1},
+ {0x1fc8, 0x1fcc, 1},
+ {0x1fd8, 0x1fdb, 1},
+ {0x1fe8, 0x1fec, 1},
+ {0x1ff8, 0x1ffc, 1},
+ {0x2126, 0x212a, 4},
+ {0x212b, 0x2132, 7},
+ {0x2183, 0x2c00, 2685},
+ {0x2c01, 0x2c2e, 1},
+ {0x2c60, 0x2c62, 2},
+ {0x2c63, 0x2c64, 1},
+ {0x2c67, 0x2c6d, 2},
+ {0x2c6e, 0x2c70, 1},
+ {0x2c72, 0x2c75, 3},
+ {0x2c7e, 0x2c80, 1},
+ {0x2c82, 0x2ce2, 2},
+ {0x2ceb, 0x2ced, 2},
+ {0x2cf2, 0xa640, 31054},
+ {0xa642, 0xa66c, 2},
+ {0xa680, 0xa69a, 2},
+ {0xa722, 0xa72e, 2},
+ {0xa732, 0xa76e, 2},
+ {0xa779, 0xa77d, 2},
+ {0xa77e, 0xa786, 2},
+ {0xa78b, 0xa78d, 2},
+ {0xa790, 0xa792, 2},
+ {0xa796, 0xa7aa, 2},
+ {0xa7ab, 0xa7ad, 1},
+ {0xa7b0, 0xa7b1, 1},
+ {0xff21, 0xff3a, 1},
+ },
+ R32: []Range32{
+ {0x10400, 0x10427, 1},
+ {0x118a0, 0x118bf, 1},
+ },
+ LatinOffset: 3,
+}
+
+var foldLt = &RangeTable{
+ R16: []Range16{
+ {0x01c4, 0x01c6, 2},
+ {0x01c7, 0x01c9, 2},
+ {0x01ca, 0x01cc, 2},
+ {0x01f1, 0x01f3, 2},
+ {0x1f80, 0x1f87, 1},
+ {0x1f90, 0x1f97, 1},
+ {0x1fa0, 0x1fa7, 1},
+ {0x1fb3, 0x1fc3, 16},
+ {0x1ff3, 0x1ff3, 1},
+ },
+}
+
+var foldLu = &RangeTable{
+ R16: []Range16{
+ {0x0061, 0x007a, 1},
+ {0x00b5, 0x00df, 42},
+ {0x00e0, 0x00f6, 1},
+ {0x00f8, 0x00ff, 1},
+ {0x0101, 0x012f, 2},
+ {0x0133, 0x0137, 2},
+ {0x013a, 0x0148, 2},
+ {0x014b, 0x0177, 2},
+ {0x017a, 0x017e, 2},
+ {0x017f, 0x0180, 1},
+ {0x0183, 0x0185, 2},
+ {0x0188, 0x018c, 4},
+ {0x0192, 0x0195, 3},
+ {0x0199, 0x019a, 1},
+ {0x019e, 0x01a1, 3},
+ {0x01a3, 0x01a5, 2},
+ {0x01a8, 0x01ad, 5},
+ {0x01b0, 0x01b4, 4},
+ {0x01b6, 0x01b9, 3},
+ {0x01bd, 0x01bf, 2},
+ {0x01c5, 0x01c6, 1},
+ {0x01c8, 0x01c9, 1},
+ {0x01cb, 0x01cc, 1},
+ {0x01ce, 0x01dc, 2},
+ {0x01dd, 0x01ef, 2},
+ {0x01f2, 0x01f3, 1},
+ {0x01f5, 0x01f9, 4},
+ {0x01fb, 0x021f, 2},
+ {0x0223, 0x0233, 2},
+ {0x023c, 0x023f, 3},
+ {0x0240, 0x0242, 2},
+ {0x0247, 0x024f, 2},
+ {0x0250, 0x0254, 1},
+ {0x0256, 0x0257, 1},
+ {0x0259, 0x025b, 2},
+ {0x025c, 0x0260, 4},
+ {0x0261, 0x0265, 2},
+ {0x0266, 0x0268, 2},
+ {0x0269, 0x026b, 2},
+ {0x026c, 0x026f, 3},
+ {0x0271, 0x0272, 1},
+ {0x0275, 0x027d, 8},
+ {0x0280, 0x0283, 3},
+ {0x0287, 0x028c, 1},
+ {0x0292, 0x029e, 12},
+ {0x0345, 0x0371, 44},
+ {0x0373, 0x037b, 4},
+ {0x037c, 0x037d, 1},
+ {0x03ac, 0x03af, 1},
+ {0x03b1, 0x03ce, 1},
+ {0x03d0, 0x03d1, 1},
+ {0x03d5, 0x03d7, 1},
+ {0x03d9, 0x03ef, 2},
+ {0x03f0, 0x03f3, 1},
+ {0x03f5, 0x03fb, 3},
+ {0x0430, 0x045f, 1},
+ {0x0461, 0x0481, 2},
+ {0x048b, 0x04bf, 2},
+ {0x04c2, 0x04ce, 2},
+ {0x04cf, 0x052f, 2},
+ {0x0561, 0x0586, 1},
+ {0x1d79, 0x1d7d, 4},
+ {0x1e01, 0x1e95, 2},
+ {0x1e9b, 0x1ea1, 6},
+ {0x1ea3, 0x1eff, 2},
+ {0x1f00, 0x1f07, 1},
+ {0x1f10, 0x1f15, 1},
+ {0x1f20, 0x1f27, 1},
+ {0x1f30, 0x1f37, 1},
+ {0x1f40, 0x1f45, 1},
+ {0x1f51, 0x1f57, 2},
+ {0x1f60, 0x1f67, 1},
+ {0x1f70, 0x1f7d, 1},
+ {0x1fb0, 0x1fb1, 1},
+ {0x1fbe, 0x1fd0, 18},
+ {0x1fd1, 0x1fe0, 15},
+ {0x1fe1, 0x1fe5, 4},
+ {0x214e, 0x2184, 54},
+ {0x2c30, 0x2c5e, 1},
+ {0x2c61, 0x2c65, 4},
+ {0x2c66, 0x2c6c, 2},
+ {0x2c73, 0x2c76, 3},
+ {0x2c81, 0x2ce3, 2},
+ {0x2cec, 0x2cee, 2},
+ {0x2cf3, 0x2d00, 13},
+ {0x2d01, 0x2d25, 1},
+ {0x2d27, 0x2d2d, 6},
+ {0xa641, 0xa66d, 2},
+ {0xa681, 0xa69b, 2},
+ {0xa723, 0xa72f, 2},
+ {0xa733, 0xa76f, 2},
+ {0xa77a, 0xa77c, 2},
+ {0xa77f, 0xa787, 2},
+ {0xa78c, 0xa791, 5},
+ {0xa793, 0xa797, 4},
+ {0xa799, 0xa7a9, 2},
+ {0xff41, 0xff5a, 1},
+ },
+ R32: []Range32{
+ {0x10428, 0x1044f, 1},
+ {0x118c0, 0x118df, 1},
+ },
+ LatinOffset: 4,
+}
+
+var foldM = &RangeTable{
+ R16: []Range16{
+ {0x0399, 0x03b9, 32},
+ {0x1fbe, 0x1fbe, 1},
+ },
+}
+
+var foldMn = &RangeTable{
+ R16: []Range16{
+ {0x0399, 0x03b9, 32},
+ {0x1fbe, 0x1fbe, 1},
+ },
+}
+
+// FoldScript maps a script name to a table of
+// code points outside the script that are equivalent under
+// simple case folding to code points inside the script.
+// If there is no entry for a script name, there are no such points.
+var FoldScript = map[string]*RangeTable{}
+
+// Range entries: 3532 16-bit, 1204 32-bit, 4736 total.
+// Range bytes: 21192 16-bit, 14448 32-bit, 35640 total.
+
+// Fold orbit bytes: 63 pairs, 252 bytes
diff --git a/src/unicode/utf16/export_test.go b/src/unicode/utf16/export_test.go
new file mode 100644
index 000000000..306247e48
--- /dev/null
+++ b/src/unicode/utf16/export_test.go
@@ -0,0 +1,11 @@
+// Copyright 2012 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package utf16
+
+// Extra names for constants so we can validate them during testing.
+const (
+ MaxRune = maxRune
+ ReplacementChar = replacementChar
+)
diff --git a/src/unicode/utf16/utf16.go b/src/unicode/utf16/utf16.go
new file mode 100644
index 000000000..c0e47c535
--- /dev/null
+++ b/src/unicode/utf16/utf16.go
@@ -0,0 +1,108 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package utf16 implements encoding and decoding of UTF-16 sequences.
+package utf16
+
+// The conditions replacementChar==unicode.ReplacementChar and
+// maxRune==unicode.MaxRune are verified in the tests.
+// Defining them locally avoids this package depending on package unicode.
+
+const (
+ replacementChar = '\uFFFD' // Unicode replacement character
+ maxRune = '\U0010FFFF' // Maximum valid Unicode code point.
+)
+
+const (
+ // 0xd800-0xdc00 encodes the high 10 bits of a pair.
+ // 0xdc00-0xe000 encodes the low 10 bits of a pair.
+ // the value is those 20 bits plus 0x10000.
+ surr1 = 0xd800
+ surr2 = 0xdc00
+ surr3 = 0xe000
+
+ surrSelf = 0x10000
+)
+
+// IsSurrogate returns true if the specified Unicode code point
+// can appear in a surrogate pair.
+func IsSurrogate(r rune) bool {
+ return surr1 <= r && r < surr3
+}
+
+// DecodeRune returns the UTF-16 decoding of a surrogate pair.
+// If the pair is not a valid UTF-16 surrogate pair, DecodeRune returns
+// the Unicode replacement code point U+FFFD.
+func DecodeRune(r1, r2 rune) rune {
+ if surr1 <= r1 && r1 < surr2 && surr2 <= r2 && r2 < surr3 {
+ return (r1-surr1)<<10 | (r2 - surr2) + 0x10000
+ }
+ return replacementChar
+}
+
+// EncodeRune returns the UTF-16 surrogate pair r1, r2 for the given rune.
+// If the rune is not a valid Unicode code point or does not need encoding,
+// EncodeRune returns U+FFFD, U+FFFD.
+func EncodeRune(r rune) (r1, r2 rune) {
+ if r < surrSelf || r > maxRune || IsSurrogate(r) {
+ return replacementChar, replacementChar
+ }
+ r -= surrSelf
+ return surr1 + (r>>10)&0x3ff, surr2 + r&0x3ff
+}
+
+// Encode returns the UTF-16 encoding of the Unicode code point sequence s.
+func Encode(s []rune) []uint16 {
+ n := len(s)
+ for _, v := range s {
+ if v >= surrSelf {
+ n++
+ }
+ }
+
+ a := make([]uint16, n)
+ n = 0
+ for _, v := range s {
+ switch {
+ case v < 0, surr1 <= v && v < surr3, v > maxRune:
+ v = replacementChar
+ fallthrough
+ case v < surrSelf:
+ a[n] = uint16(v)
+ n++
+ default:
+ r1, r2 := EncodeRune(v)
+ a[n] = uint16(r1)
+ a[n+1] = uint16(r2)
+ n += 2
+ }
+ }
+ return a[0:n]
+}
+
+// Decode returns the Unicode code point sequence represented
+// by the UTF-16 encoding s.
+func Decode(s []uint16) []rune {
+ a := make([]rune, len(s))
+ n := 0
+ for i := 0; i < len(s); i++ {
+ switch r := s[i]; {
+ case surr1 <= r && r < surr2 && i+1 < len(s) &&
+ surr2 <= s[i+1] && s[i+1] < surr3:
+ // valid surrogate sequence
+ a[n] = DecodeRune(rune(r), rune(s[i+1]))
+ i++
+ n++
+ case surr1 <= r && r < surr3:
+ // invalid surrogate sequence
+ a[n] = replacementChar
+ n++
+ default:
+ // normal rune
+ a[n] = rune(r)
+ n++
+ }
+ }
+ return a[0:n]
+}
diff --git a/src/unicode/utf16/utf16_test.go b/src/unicode/utf16/utf16_test.go
new file mode 100644
index 000000000..3dca472bb
--- /dev/null
+++ b/src/unicode/utf16/utf16_test.go
@@ -0,0 +1,149 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package utf16_test
+
+import (
+ "reflect"
+ "testing"
+ "unicode"
+ . "unicode/utf16"
+)
+
+// Validate the constants redefined from unicode.
+func TestConstants(t *testing.T) {
+ if MaxRune != unicode.MaxRune {
+ t.Errorf("utf16.maxRune is wrong: %x should be %x", MaxRune, unicode.MaxRune)
+ }
+ if ReplacementChar != unicode.ReplacementChar {
+ t.Errorf("utf16.replacementChar is wrong: %x should be %x", ReplacementChar, unicode.ReplacementChar)
+ }
+}
+
+type encodeTest struct {
+ in []rune
+ out []uint16
+}
+
+var encodeTests = []encodeTest{
+ {[]rune{1, 2, 3, 4}, []uint16{1, 2, 3, 4}},
+ {[]rune{0xffff, 0x10000, 0x10001, 0x12345, 0x10ffff},
+ []uint16{0xffff, 0xd800, 0xdc00, 0xd800, 0xdc01, 0xd808, 0xdf45, 0xdbff, 0xdfff}},
+ {[]rune{'a', 'b', 0xd7ff, 0xd800, 0xdfff, 0xe000, 0x110000, -1},
+ []uint16{'a', 'b', 0xd7ff, 0xfffd, 0xfffd, 0xe000, 0xfffd, 0xfffd}},
+}
+
+func TestEncode(t *testing.T) {
+ for _, tt := range encodeTests {
+ out := Encode(tt.in)
+ if !reflect.DeepEqual(out, tt.out) {
+ t.Errorf("Encode(%x) = %x; want %x", tt.in, out, tt.out)
+ }
+ }
+}
+
+func TestEncodeRune(t *testing.T) {
+ for i, tt := range encodeTests {
+ j := 0
+ for _, r := range tt.in {
+ r1, r2 := EncodeRune(r)
+ if r < 0x10000 || r > unicode.MaxRune {
+ if j >= len(tt.out) {
+ t.Errorf("#%d: ran out of tt.out", i)
+ break
+ }
+ if r1 != unicode.ReplacementChar || r2 != unicode.ReplacementChar {
+ t.Errorf("EncodeRune(%#x) = %#x, %#x; want 0xfffd, 0xfffd", r, r1, r2)
+ }
+ j++
+ } else {
+ if j+1 >= len(tt.out) {
+ t.Errorf("#%d: ran out of tt.out", i)
+ break
+ }
+ if r1 != rune(tt.out[j]) || r2 != rune(tt.out[j+1]) {
+ t.Errorf("EncodeRune(%#x) = %#x, %#x; want %#x, %#x", r, r1, r2, tt.out[j], tt.out[j+1])
+ }
+ j += 2
+ dec := DecodeRune(r1, r2)
+ if dec != r {
+ t.Errorf("DecodeRune(%#x, %#x) = %#x; want %#x", r1, r2, dec, r)
+ }
+ }
+ }
+ if j != len(tt.out) {
+ t.Errorf("#%d: EncodeRune didn't generate enough output", i)
+ }
+ }
+}
+
+type decodeTest struct {
+ in []uint16
+ out []rune
+}
+
+var decodeTests = []decodeTest{
+ {[]uint16{1, 2, 3, 4}, []rune{1, 2, 3, 4}},
+ {[]uint16{0xffff, 0xd800, 0xdc00, 0xd800, 0xdc01, 0xd808, 0xdf45, 0xdbff, 0xdfff},
+ []rune{0xffff, 0x10000, 0x10001, 0x12345, 0x10ffff}},
+ {[]uint16{0xd800, 'a'}, []rune{0xfffd, 'a'}},
+ {[]uint16{0xdfff}, []rune{0xfffd}},
+}
+
+func TestDecode(t *testing.T) {
+ for _, tt := range decodeTests {
+ out := Decode(tt.in)
+ if !reflect.DeepEqual(out, tt.out) {
+ t.Errorf("Decode(%x) = %x; want %x", tt.in, out, tt.out)
+ }
+ }
+}
+
+var decodeRuneTests = []struct {
+ r1, r2 rune
+ want rune
+}{
+ {0xd800, 0xdc00, 0x10000},
+ {0xd800, 0xdc01, 0x10001},
+ {0xd808, 0xdf45, 0x12345},
+ {0xdbff, 0xdfff, 0x10ffff},
+ {0xd800, 'a', 0xfffd}, // illegal, replacement rune substituted
+}
+
+func TestDecodeRune(t *testing.T) {
+ for i, tt := range decodeRuneTests {
+ got := DecodeRune(tt.r1, tt.r2)
+ if got != tt.want {
+ t.Errorf("%d: DecodeRune(%q, %q) = %v; want %v", i, tt.r1, tt.r2, got, tt.want)
+ }
+ }
+}
+
+var surrogateTests = []struct {
+ r rune
+ want bool
+}{
+ // from http://en.wikipedia.org/wiki/UTF-16
+ {'\u007A', false}, // LATIN SMALL LETTER Z
+ {'\u6C34', false}, // CJK UNIFIED IDEOGRAPH-6C34 (water)
+ {'\uFEFF', false}, // Byte Order Mark
+ {'\U00010000', false}, // LINEAR B SYLLABLE B008 A (first non-BMP code point)
+ {'\U0001D11E', false}, // MUSICAL SYMBOL G CLEF
+ {'\U0010FFFD', false}, // PRIVATE USE CHARACTER-10FFFD (last Unicode code point)
+
+ {rune(0xd7ff), false}, // surr1-1
+ {rune(0xd800), true}, // surr1
+ {rune(0xdc00), true}, // surr2
+ {rune(0xe000), false}, // surr3
+ {rune(0xdfff), true}, // surr3-1
+}
+
+func TestIsSurrogate(t *testing.T) {
+ for i, tt := range surrogateTests {
+ got := IsSurrogate(tt.r)
+ if got != tt.want {
+ t.Errorf("%d: IsSurrogate(%q) = %v; want %v", i, tt.r, got, tt.want)
+ }
+ }
+}
diff --git a/src/unicode/utf8/example_test.go b/src/unicode/utf8/example_test.go
new file mode 100644
index 000000000..7b3e7ac74
--- /dev/null
+++ b/src/unicode/utf8/example_test.go
@@ -0,0 +1,196 @@
+// Copyright 2013 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package utf8_test
+
+import (
+ "fmt"
+ "unicode/utf8"
+)
+
+func ExampleDecodeLastRune() {
+ b := []byte("Hello, 世界")
+
+ for len(b) > 0 {
+ r, size := utf8.DecodeLastRune(b)
+ fmt.Printf("%c %v\n", r, size)
+
+ b = b[:len(b)-size]
+ }
+ // Output:
+ // 界 3
+ // 世 3
+ // 1
+ // , 1
+ // o 1
+ // l 1
+ // l 1
+ // e 1
+ // H 1
+}
+
+func ExampleDecodeLastRuneInString() {
+ str := "Hello, 世界"
+
+ for len(str) > 0 {
+ r, size := utf8.DecodeLastRuneInString(str)
+ fmt.Printf("%c %v\n", r, size)
+
+ str = str[:len(str)-size]
+ }
+ // Output:
+ // 界 3
+ // 世 3
+ // 1
+ // , 1
+ // o 1
+ // l 1
+ // l 1
+ // e 1
+ // H 1
+
+}
+
+func ExampleDecodeRune() {
+ b := []byte("Hello, 世界")
+
+ for len(b) > 0 {
+ r, size := utf8.DecodeRune(b)
+ fmt.Printf("%c %v\n", r, size)
+
+ b = b[size:]
+ }
+ // Output:
+ // H 1
+ // e 1
+ // l 1
+ // l 1
+ // o 1
+ // , 1
+ // 1
+ // 世 3
+ // 界 3
+}
+
+func ExampleDecodeRuneInString() {
+ str := "Hello, 世界"
+
+ for len(str) > 0 {
+ r, size := utf8.DecodeRuneInString(str)
+ fmt.Printf("%c %v\n", r, size)
+
+ str = str[size:]
+ }
+ // Output:
+ // H 1
+ // e 1
+ // l 1
+ // l 1
+ // o 1
+ // , 1
+ // 1
+ // 世 3
+ // 界 3
+}
+
+func ExampleEncodeRune() {
+ r := '世'
+ buf := make([]byte, 3)
+
+ n := utf8.EncodeRune(buf, r)
+
+ fmt.Println(buf)
+ fmt.Println(n)
+ // Output:
+ // [228 184 150]
+ // 3
+}
+
+func ExampleFullRune() {
+ buf := []byte{228, 184, 150} // 世
+ fmt.Println(utf8.FullRune(buf))
+ fmt.Println(utf8.FullRune(buf[:2]))
+ // Output:
+ // true
+ // false
+}
+
+func ExampleFullRuneInString() {
+ str := "世"
+ fmt.Println(utf8.FullRuneInString(str))
+ fmt.Println(utf8.FullRuneInString(str[:2]))
+ // Output:
+ // true
+ // false
+}
+
+func ExampleRuneCount() {
+ buf := []byte("Hello, 世界")
+ fmt.Println("bytes =", len(buf))
+ fmt.Println("runes =", utf8.RuneCount(buf))
+ // Output:
+ // bytes = 13
+ // runes = 9
+}
+
+func ExampleRuneCountInString() {
+ str := "Hello, 世界"
+ fmt.Println("bytes =", len(str))
+ fmt.Println("runes =", utf8.RuneCountInString(str))
+ // Output:
+ // bytes = 13
+ // runes = 9
+}
+
+func ExampleRuneLen() {
+ fmt.Println(utf8.RuneLen('a'))
+ fmt.Println(utf8.RuneLen('界'))
+ // Output:
+ // 1
+ // 3
+}
+
+func ExampleRuneStart() {
+ buf := []byte("a界")
+ fmt.Println(utf8.RuneStart(buf[0]))
+ fmt.Println(utf8.RuneStart(buf[1]))
+ fmt.Println(utf8.RuneStart(buf[2]))
+ // Output:
+ // true
+ // true
+ // false
+}
+
+func ExampleValid() {
+ valid := []byte("Hello, 世界")
+ invalid := []byte{0xff, 0xfe, 0xfd}
+
+ fmt.Println(utf8.Valid(valid))
+ fmt.Println(utf8.Valid(invalid))
+ // Output:
+ // true
+ // false
+}
+
+func ExampleValidRune() {
+ valid := 'a'
+ invalid := rune(0xfffffff)
+
+ fmt.Println(utf8.ValidRune(valid))
+ fmt.Println(utf8.ValidRune(invalid))
+ // Output:
+ // true
+ // false
+}
+
+func ExampleValidString() {
+ valid := "Hello, 世界"
+ invalid := string([]byte{0xff, 0xfe, 0xfd})
+
+ fmt.Println(utf8.ValidString(valid))
+ fmt.Println(utf8.ValidString(invalid))
+ // Output:
+ // true
+ // false
+}
diff --git a/src/unicode/utf8/utf8.go b/src/unicode/utf8/utf8.go
new file mode 100644
index 000000000..253295ad3
--- /dev/null
+++ b/src/unicode/utf8/utf8.go
@@ -0,0 +1,435 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package utf8 implements functions and constants to support text encoded in
+// UTF-8. It includes functions to translate between runes and UTF-8 byte sequences.
+package utf8
+
+// The conditions RuneError==unicode.ReplacementChar and
+// MaxRune==unicode.MaxRune are verified in the tests.
+// Defining them locally avoids this package depending on package unicode.
+
+// Numbers fundamental to the encoding.
+const (
+ RuneError = '\uFFFD' // the "error" Rune or "Unicode replacement character"
+ RuneSelf = 0x80 // characters below Runeself are represented as themselves in a single byte.
+ MaxRune = '\U0010FFFF' // Maximum valid Unicode code point.
+ UTFMax = 4 // maximum number of bytes of a UTF-8 encoded Unicode character.
+)
+
+// Code points in the surrogate range are not valid for UTF-8.
+const (
+ surrogateMin = 0xD800
+ surrogateMax = 0xDFFF
+)
+
+const (
+ t1 = 0x00 // 0000 0000
+ tx = 0x80 // 1000 0000
+ t2 = 0xC0 // 1100 0000
+ t3 = 0xE0 // 1110 0000
+ t4 = 0xF0 // 1111 0000
+ t5 = 0xF8 // 1111 1000
+
+ maskx = 0x3F // 0011 1111
+ mask2 = 0x1F // 0001 1111
+ mask3 = 0x0F // 0000 1111
+ mask4 = 0x07 // 0000 0111
+
+ rune1Max = 1<<7 - 1
+ rune2Max = 1<<11 - 1
+ rune3Max = 1<<16 - 1
+)
+
+func decodeRuneInternal(p []byte) (r rune, size int, short bool) {
+ n := len(p)
+ if n < 1 {
+ return RuneError, 0, true
+ }
+ c0 := p[0]
+
+ // 1-byte, 7-bit sequence?
+ if c0 < tx {
+ return rune(c0), 1, false
+ }
+
+ // unexpected continuation byte?
+ if c0 < t2 {
+ return RuneError, 1, false
+ }
+
+ // need first continuation byte
+ if n < 2 {
+ return RuneError, 1, true
+ }
+ c1 := p[1]
+ if c1 < tx || t2 <= c1 {
+ return RuneError, 1, false
+ }
+
+ // 2-byte, 11-bit sequence?
+ if c0 < t3 {
+ r = rune(c0&mask2)<<6 | rune(c1&maskx)
+ if r <= rune1Max {
+ return RuneError, 1, false
+ }
+ return r, 2, false
+ }
+
+ // need second continuation byte
+ if n < 3 {
+ return RuneError, 1, true
+ }
+ c2 := p[2]
+ if c2 < tx || t2 <= c2 {
+ return RuneError, 1, false
+ }
+
+ // 3-byte, 16-bit sequence?
+ if c0 < t4 {
+ r = rune(c0&mask3)<<12 | rune(c1&maskx)<<6 | rune(c2&maskx)
+ if r <= rune2Max {
+ return RuneError, 1, false
+ }
+ if surrogateMin <= r && r <= surrogateMax {
+ return RuneError, 1, false
+ }
+ return r, 3, false
+ }
+
+ // need third continuation byte
+ if n < 4 {
+ return RuneError, 1, true
+ }
+ c3 := p[3]
+ if c3 < tx || t2 <= c3 {
+ return RuneError, 1, false
+ }
+
+ // 4-byte, 21-bit sequence?
+ if c0 < t5 {
+ r = rune(c0&mask4)<<18 | rune(c1&maskx)<<12 | rune(c2&maskx)<<6 | rune(c3&maskx)
+ if r <= rune3Max || MaxRune < r {
+ return RuneError, 1, false
+ }
+ return r, 4, false
+ }
+
+ // error
+ return RuneError, 1, false
+}
+
+func decodeRuneInStringInternal(s string) (r rune, size int, short bool) {
+ n := len(s)
+ if n < 1 {
+ return RuneError, 0, true
+ }
+ c0 := s[0]
+
+ // 1-byte, 7-bit sequence?
+ if c0 < tx {
+ return rune(c0), 1, false
+ }
+
+ // unexpected continuation byte?
+ if c0 < t2 {
+ return RuneError, 1, false
+ }
+
+ // need first continuation byte
+ if n < 2 {
+ return RuneError, 1, true
+ }
+ c1 := s[1]
+ if c1 < tx || t2 <= c1 {
+ return RuneError, 1, false
+ }
+
+ // 2-byte, 11-bit sequence?
+ if c0 < t3 {
+ r = rune(c0&mask2)<<6 | rune(c1&maskx)
+ if r <= rune1Max {
+ return RuneError, 1, false
+ }
+ return r, 2, false
+ }
+
+ // need second continuation byte
+ if n < 3 {
+ return RuneError, 1, true
+ }
+ c2 := s[2]
+ if c2 < tx || t2 <= c2 {
+ return RuneError, 1, false
+ }
+
+ // 3-byte, 16-bit sequence?
+ if c0 < t4 {
+ r = rune(c0&mask3)<<12 | rune(c1&maskx)<<6 | rune(c2&maskx)
+ if r <= rune2Max {
+ return RuneError, 1, false
+ }
+ if surrogateMin <= r && r <= surrogateMax {
+ return RuneError, 1, false
+ }
+ return r, 3, false
+ }
+
+ // need third continuation byte
+ if n < 4 {
+ return RuneError, 1, true
+ }
+ c3 := s[3]
+ if c3 < tx || t2 <= c3 {
+ return RuneError, 1, false
+ }
+
+ // 4-byte, 21-bit sequence?
+ if c0 < t5 {
+ r = rune(c0&mask4)<<18 | rune(c1&maskx)<<12 | rune(c2&maskx)<<6 | rune(c3&maskx)
+ if r <= rune3Max || MaxRune < r {
+ return RuneError, 1, false
+ }
+ return r, 4, false
+ }
+
+ // error
+ return RuneError, 1, false
+}
+
+// FullRune reports whether the bytes in p begin with a full UTF-8 encoding of a rune.
+// An invalid encoding is considered a full Rune since it will convert as a width-1 error rune.
+func FullRune(p []byte) bool {
+ _, _, short := decodeRuneInternal(p)
+ return !short
+}
+
+// FullRuneInString is like FullRune but its input is a string.
+func FullRuneInString(s string) bool {
+ _, _, short := decodeRuneInStringInternal(s)
+ return !short
+}
+
+// DecodeRune unpacks the first UTF-8 encoding in p and returns the rune and its width in bytes.
+// If the encoding is invalid, it returns (RuneError, 1), an impossible result for correct UTF-8.
+// An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
+// out of range, or is not the shortest possible UTF-8 encoding for the
+// value. No other validation is performed.
+func DecodeRune(p []byte) (r rune, size int) {
+ r, size, _ = decodeRuneInternal(p)
+ return
+}
+
+// DecodeRuneInString is like DecodeRune but its input is a string.
+// If the encoding is invalid, it returns (RuneError, 1), an impossible result for correct UTF-8.
+// An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
+// out of range, or is not the shortest possible UTF-8 encoding for the
+// value. No other validation is performed.
+func DecodeRuneInString(s string) (r rune, size int) {
+ r, size, _ = decodeRuneInStringInternal(s)
+ return
+}
+
+// DecodeLastRune unpacks the last UTF-8 encoding in p and returns the rune and its width in bytes.
+// If the encoding is invalid, it returns (RuneError, 1), an impossible result for correct UTF-8.
+// An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
+// out of range, or is not the shortest possible UTF-8 encoding for the
+// value. No other validation is performed.
+func DecodeLastRune(p []byte) (r rune, size int) {
+ end := len(p)
+ if end == 0 {
+ return RuneError, 0
+ }
+ start := end - 1
+ r = rune(p[start])
+ if r < RuneSelf {
+ return r, 1
+ }
+ // guard against O(n^2) behavior when traversing
+ // backwards through strings with long sequences of
+ // invalid UTF-8.
+ lim := end - UTFMax
+ if lim < 0 {
+ lim = 0
+ }
+ for start--; start >= lim; start-- {
+ if RuneStart(p[start]) {
+ break
+ }
+ }
+ if start < 0 {
+ start = 0
+ }
+ r, size = DecodeRune(p[start:end])
+ if start+size != end {
+ return RuneError, 1
+ }
+ return r, size
+}
+
+// DecodeLastRuneInString is like DecodeLastRune but its input is a string.
+// If the encoding is invalid, it returns (RuneError, 1), an impossible result for correct UTF-8.
+// An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
+// out of range, or is not the shortest possible UTF-8 encoding for the
+// value. No other validation is performed.
+func DecodeLastRuneInString(s string) (r rune, size int) {
+ end := len(s)
+ if end == 0 {
+ return RuneError, 0
+ }
+ start := end - 1
+ r = rune(s[start])
+ if r < RuneSelf {
+ return r, 1
+ }
+ // guard against O(n^2) behavior when traversing
+ // backwards through strings with long sequences of
+ // invalid UTF-8.
+ lim := end - UTFMax
+ if lim < 0 {
+ lim = 0
+ }
+ for start--; start >= lim; start-- {
+ if RuneStart(s[start]) {
+ break
+ }
+ }
+ if start < 0 {
+ start = 0
+ }
+ r, size = DecodeRuneInString(s[start:end])
+ if start+size != end {
+ return RuneError, 1
+ }
+ return r, size
+}
+
+// RuneLen returns the number of bytes required to encode the rune.
+// It returns -1 if the rune is not a valid value to encode in UTF-8.
+func RuneLen(r rune) int {
+ switch {
+ case r < 0:
+ return -1
+ case r <= rune1Max:
+ return 1
+ case r <= rune2Max:
+ return 2
+ case surrogateMin <= r && r <= surrogateMax:
+ return -1
+ case r <= rune3Max:
+ return 3
+ case r <= MaxRune:
+ return 4
+ }
+ return -1
+}
+
+// EncodeRune writes into p (which must be large enough) the UTF-8 encoding of the rune.
+// It returns the number of bytes written.
+func EncodeRune(p []byte, r rune) int {
+ // Negative values are erroneous. Making it unsigned addresses the problem.
+ switch i := uint32(r); {
+ case i <= rune1Max:
+ p[0] = byte(r)
+ return 1
+ case i <= rune2Max:
+ p[0] = t2 | byte(r>>6)
+ p[1] = tx | byte(r)&maskx
+ return 2
+ case i > MaxRune, surrogateMin <= i && i <= surrogateMax:
+ r = RuneError
+ fallthrough
+ case i <= rune3Max:
+ p[0] = t3 | byte(r>>12)
+ p[1] = tx | byte(r>>6)&maskx
+ p[2] = tx | byte(r)&maskx
+ return 3
+ default:
+ p[0] = t4 | byte(r>>18)
+ p[1] = tx | byte(r>>12)&maskx
+ p[2] = tx | byte(r>>6)&maskx
+ p[3] = tx | byte(r)&maskx
+ return 4
+ }
+}
+
+// RuneCount returns the number of runes in p. Erroneous and short
+// encodings are treated as single runes of width 1 byte.
+func RuneCount(p []byte) int {
+ i := 0
+ var n int
+ for n = 0; i < len(p); n++ {
+ if p[i] < RuneSelf {
+ i++
+ } else {
+ _, size := DecodeRune(p[i:])
+ i += size
+ }
+ }
+ return n
+}
+
+// RuneCountInString is like RuneCount but its input is a string.
+func RuneCountInString(s string) (n int) {
+ for range s {
+ n++
+ }
+ return
+}
+
+// RuneStart reports whether the byte could be the first byte of
+// an encoded rune. Second and subsequent bytes always have the top
+// two bits set to 10.
+func RuneStart(b byte) bool { return b&0xC0 != 0x80 }
+
+// Valid reports whether p consists entirely of valid UTF-8-encoded runes.
+func Valid(p []byte) bool {
+ i := 0
+ for i < len(p) {
+ if p[i] < RuneSelf {
+ i++
+ } else {
+ _, size := DecodeRune(p[i:])
+ if size == 1 {
+ // All valid runes of size 1 (those
+ // below RuneSelf) were handled above.
+ // This must be a RuneError.
+ return false
+ }
+ i += size
+ }
+ }
+ return true
+}
+
+// ValidString reports whether s consists entirely of valid UTF-8-encoded runes.
+func ValidString(s string) bool {
+ for i, r := range s {
+ if r == RuneError {
+ // The RuneError value can be an error
+ // sentinel value (if it's size 1) or the same
+ // value encoded properly. Decode it to see if
+ // it's the 1 byte sentinel value.
+ _, size := DecodeRuneInString(s[i:])
+ if size == 1 {
+ return false
+ }
+ }
+ }
+ return true
+}
+
+// ValidRune reports whether r can be legally encoded as UTF-8.
+// Code points that are out of range or a surrogate half are illegal.
+func ValidRune(r rune) bool {
+ switch {
+ case r < 0:
+ return false
+ case surrogateMin <= r && r <= surrogateMax:
+ return false
+ case r > MaxRune:
+ return false
+ }
+ return true
+}
diff --git a/src/unicode/utf8/utf8_test.go b/src/unicode/utf8/utf8_test.go
new file mode 100644
index 000000000..758d7a0f8
--- /dev/null
+++ b/src/unicode/utf8/utf8_test.go
@@ -0,0 +1,444 @@
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package utf8_test
+
+import (
+ "bytes"
+ "testing"
+ "unicode"
+ . "unicode/utf8"
+)
+
+// Validate the constants redefined from unicode.
+func init() {
+ if MaxRune != unicode.MaxRune {
+ panic("utf8.MaxRune is wrong")
+ }
+ if RuneError != unicode.ReplacementChar {
+ panic("utf8.RuneError is wrong")
+ }
+}
+
+// Validate the constants redefined from unicode.
+func TestConstants(t *testing.T) {
+ if MaxRune != unicode.MaxRune {
+ t.Errorf("utf8.MaxRune is wrong: %x should be %x", MaxRune, unicode.MaxRune)
+ }
+ if RuneError != unicode.ReplacementChar {
+ t.Errorf("utf8.RuneError is wrong: %x should be %x", RuneError, unicode.ReplacementChar)
+ }
+}
+
+type Utf8Map struct {
+ r rune
+ str string
+}
+
+var utf8map = []Utf8Map{
+ {0x0000, "\x00"},
+ {0x0001, "\x01"},
+ {0x007e, "\x7e"},
+ {0x007f, "\x7f"},
+ {0x0080, "\xc2\x80"},
+ {0x0081, "\xc2\x81"},
+ {0x00bf, "\xc2\xbf"},
+ {0x00c0, "\xc3\x80"},
+ {0x00c1, "\xc3\x81"},
+ {0x00c8, "\xc3\x88"},
+ {0x00d0, "\xc3\x90"},
+ {0x00e0, "\xc3\xa0"},
+ {0x00f0, "\xc3\xb0"},
+ {0x00f8, "\xc3\xb8"},
+ {0x00ff, "\xc3\xbf"},
+ {0x0100, "\xc4\x80"},
+ {0x07ff, "\xdf\xbf"},
+ {0x0800, "\xe0\xa0\x80"},
+ {0x0801, "\xe0\xa0\x81"},
+ {0xd7ff, "\xed\x9f\xbf"}, // last code point before surrogate half.
+ {0xe000, "\xee\x80\x80"}, // first code point after surrogate half.
+ {0xfffe, "\xef\xbf\xbe"},
+ {0xffff, "\xef\xbf\xbf"},
+ {0x10000, "\xf0\x90\x80\x80"},
+ {0x10001, "\xf0\x90\x80\x81"},
+ {0x10fffe, "\xf4\x8f\xbf\xbe"},
+ {0x10ffff, "\xf4\x8f\xbf\xbf"},
+ {0xFFFD, "\xef\xbf\xbd"},
+}
+
+var surrogateMap = []Utf8Map{
+ {0xd800, "\xed\xa0\x80"}, // surrogate min decodes to (RuneError, 1)
+ {0xdfff, "\xed\xbf\xbf"}, // surrogate max decodes to (RuneError, 1)
+}
+
+var testStrings = []string{
+ "",
+ "abcd",
+ "☺☻☹",
+ "日a本b語ç日ð本Ê語þ日¥本¼語i日©",
+ "日a本b語ç日ð本Ê語þ日¥本¼語i日©日a本b語ç日ð本Ê語þ日¥本¼語i日©日a本b語ç日ð本Ê語þ日¥本¼語i日©",
+ "\x80\x80\x80\x80",
+}
+
+func TestFullRune(t *testing.T) {
+ for _, m := range utf8map {
+ b := []byte(m.str)
+ if !FullRune(b) {
+ t.Errorf("FullRune(%q) (%U) = false, want true", b, m.r)
+ }
+ s := m.str
+ if !FullRuneInString(s) {
+ t.Errorf("FullRuneInString(%q) (%U) = false, want true", s, m.r)
+ }
+ b1 := b[0 : len(b)-1]
+ if FullRune(b1) {
+ t.Errorf("FullRune(%q) = true, want false", b1)
+ }
+ s1 := string(b1)
+ if FullRuneInString(s1) {
+ t.Errorf("FullRune(%q) = true, want false", s1)
+ }
+ }
+}
+
+func TestEncodeRune(t *testing.T) {
+ for _, m := range utf8map {
+ b := []byte(m.str)
+ var buf [10]byte
+ n := EncodeRune(buf[0:], m.r)
+ b1 := buf[0:n]
+ if !bytes.Equal(b, b1) {
+ t.Errorf("EncodeRune(%#04x) = %q want %q", m.r, b1, b)
+ }
+ }
+}
+
+func TestDecodeRune(t *testing.T) {
+ for _, m := range utf8map {
+ b := []byte(m.str)
+ r, size := DecodeRune(b)
+ if r != m.r || size != len(b) {
+ t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
+ }
+ s := m.str
+ r, size = DecodeRuneInString(s)
+ if r != m.r || size != len(b) {
+ t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
+ }
+
+ // there's an extra byte that bytes left behind - make sure trailing byte works
+ r, size = DecodeRune(b[0:cap(b)])
+ if r != m.r || size != len(b) {
+ t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
+ }
+ s = m.str + "\x00"
+ r, size = DecodeRuneInString(s)
+ if r != m.r || size != len(b) {
+ t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
+ }
+
+ // make sure missing bytes fail
+ wantsize := 1
+ if wantsize >= len(b) {
+ wantsize = 0
+ }
+ r, size = DecodeRune(b[0 : len(b)-1])
+ if r != RuneError || size != wantsize {
+ t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b[0:len(b)-1], r, size, RuneError, wantsize)
+ }
+ s = m.str[0 : len(m.str)-1]
+ r, size = DecodeRuneInString(s)
+ if r != RuneError || size != wantsize {
+ t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, wantsize)
+ }
+
+ // make sure bad sequences fail
+ if len(b) == 1 {
+ b[0] = 0x80
+ } else {
+ b[len(b)-1] = 0x7F
+ }
+ r, size = DecodeRune(b)
+ if r != RuneError || size != 1 {
+ t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, RuneError, 1)
+ }
+ s = string(b)
+ r, size = DecodeRuneInString(s)
+ if r != RuneError || size != 1 {
+ t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, 1)
+ }
+
+ }
+}
+
+func TestDecodeSurrogateRune(t *testing.T) {
+ for _, m := range surrogateMap {
+ b := []byte(m.str)
+ r, size := DecodeRune(b)
+ if r != RuneError || size != 1 {
+ t.Errorf("DecodeRune(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
+ }
+ s := m.str
+ r, size = DecodeRuneInString(s)
+ if r != RuneError || size != 1 {
+ t.Errorf("DecodeRuneInString(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
+ }
+ }
+}
+
+// Check that DecodeRune and DecodeLastRune correspond to
+// the equivalent range loop.
+func TestSequencing(t *testing.T) {
+ for _, ts := range testStrings {
+ for _, m := range utf8map {
+ for _, s := range []string{ts + m.str, m.str + ts, ts + m.str + ts} {
+ testSequence(t, s)
+ }
+ }
+ }
+}
+
+// Check that a range loop and a []int conversion visit the same runes.
+// Not really a test of this package, but the assumption is used here and
+// it's good to verify
+func TestIntConversion(t *testing.T) {
+ for _, ts := range testStrings {
+ runes := []rune(ts)
+ if RuneCountInString(ts) != len(runes) {
+ t.Errorf("%q: expected %d runes; got %d", ts, len(runes), RuneCountInString(ts))
+ break
+ }
+ i := 0
+ for _, r := range ts {
+ if r != runes[i] {
+ t.Errorf("%q[%d]: expected %c (%U); got %c (%U)", ts, i, runes[i], runes[i], r, r)
+ }
+ i++
+ }
+ }
+}
+
+func testSequence(t *testing.T, s string) {
+ type info struct {
+ index int
+ r rune
+ }
+ index := make([]info, len(s))
+ b := []byte(s)
+ si := 0
+ j := 0
+ for i, r := range s {
+ if si != i {
+ t.Errorf("Sequence(%q) mismatched index %d, want %d", s, si, i)
+ return
+ }
+ index[j] = info{i, r}
+ j++
+ r1, size1 := DecodeRune(b[i:])
+ if r != r1 {
+ t.Errorf("DecodeRune(%q) = %#04x, want %#04x", s[i:], r1, r)
+ return
+ }
+ r2, size2 := DecodeRuneInString(s[i:])
+ if r != r2 {
+ t.Errorf("DecodeRuneInString(%q) = %#04x, want %#04x", s[i:], r2, r)
+ return
+ }
+ if size1 != size2 {
+ t.Errorf("DecodeRune/DecodeRuneInString(%q) size mismatch %d/%d", s[i:], size1, size2)
+ return
+ }
+ si += size1
+ }
+ j--
+ for si = len(s); si > 0; {
+ r1, size1 := DecodeLastRune(b[0:si])
+ r2, size2 := DecodeLastRuneInString(s[0:si])
+ if size1 != size2 {
+ t.Errorf("DecodeLastRune/DecodeLastRuneInString(%q, %d) size mismatch %d/%d", s, si, size1, size2)
+ return
+ }
+ if r1 != index[j].r {
+ t.Errorf("DecodeLastRune(%q, %d) = %#04x, want %#04x", s, si, r1, index[j].r)
+ return
+ }
+ if r2 != index[j].r {
+ t.Errorf("DecodeLastRuneInString(%q, %d) = %#04x, want %#04x", s, si, r2, index[j].r)
+ return
+ }
+ si -= size1
+ if si != index[j].index {
+ t.Errorf("DecodeLastRune(%q) index mismatch at %d, want %d", s, si, index[j].index)
+ return
+ }
+ j--
+ }
+ if si != 0 {
+ t.Errorf("DecodeLastRune(%q) finished at %d, not 0", s, si)
+ }
+}
+
+// Check that negative runes encode as U+FFFD.
+func TestNegativeRune(t *testing.T) {
+ errorbuf := make([]byte, UTFMax)
+ errorbuf = errorbuf[0:EncodeRune(errorbuf, RuneError)]
+ buf := make([]byte, UTFMax)
+ buf = buf[0:EncodeRune(buf, -1)]
+ if !bytes.Equal(buf, errorbuf) {
+ t.Errorf("incorrect encoding [% x] for -1; expected [% x]", buf, errorbuf)
+ }
+}
+
+type RuneCountTest struct {
+ in string
+ out int
+}
+
+var runecounttests = []RuneCountTest{
+ {"abcd", 4},
+ {"☺☻☹", 3},
+ {"1,2,3,4", 7},
+ {"\xe2\x00", 2},
+}
+
+func TestRuneCount(t *testing.T) {
+ for _, tt := range runecounttests {
+ if out := RuneCountInString(tt.in); out != tt.out {
+ t.Errorf("RuneCountInString(%q) = %d, want %d", tt.in, out, tt.out)
+ }
+ if out := RuneCount([]byte(tt.in)); out != tt.out {
+ t.Errorf("RuneCount(%q) = %d, want %d", tt.in, out, tt.out)
+ }
+ }
+}
+
+type RuneLenTest struct {
+ r rune
+ size int
+}
+
+var runelentests = []RuneLenTest{
+ {0, 1},
+ {'e', 1},
+ {'é', 2},
+ {'☺', 3},
+ {RuneError, 3},
+ {MaxRune, 4},
+ {0xD800, -1},
+ {0xDFFF, -1},
+ {MaxRune + 1, -1},
+ {-1, -1},
+}
+
+func TestRuneLen(t *testing.T) {
+ for _, tt := range runelentests {
+ if size := RuneLen(tt.r); size != tt.size {
+ t.Errorf("RuneLen(%#U) = %d, want %d", tt.r, size, tt.size)
+ }
+ }
+}
+
+type ValidTest struct {
+ in string
+ out bool
+}
+
+var validTests = []ValidTest{
+ {"", true},
+ {"a", true},
+ {"abc", true},
+ {"Ж", true},
+ {"ЖЖ", true},
+ {"брэд-ЛГТМ", true},
+ {"☺☻☹", true},
+ {string([]byte{66, 250}), false},
+ {string([]byte{66, 250, 67}), false},
+ {"a\uFFFDb", true},
+ {string("\xF4\x8F\xBF\xBF"), true}, // U+10FFFF
+ {string("\xF4\x90\x80\x80"), false}, // U+10FFFF+1; out of range
+ {string("\xF7\xBF\xBF\xBF"), false}, // 0x1FFFFF; out of range
+ {string("\xFB\xBF\xBF\xBF\xBF"), false}, // 0x3FFFFFF; out of range
+ {string("\xc0\x80"), false}, // U+0000 encoded in two bytes: incorrect
+ {string("\xed\xa0\x80"), false}, // U+D800 high surrogate (sic)
+ {string("\xed\xbf\xbf"), false}, // U+DFFF low surrogate (sic)
+}
+
+func TestValid(t *testing.T) {
+ for _, tt := range validTests {
+ if Valid([]byte(tt.in)) != tt.out {
+ t.Errorf("Valid(%q) = %v; want %v", tt.in, !tt.out, tt.out)
+ }
+ if ValidString(tt.in) != tt.out {
+ t.Errorf("ValidString(%q) = %v; want %v", tt.in, !tt.out, tt.out)
+ }
+ }
+}
+
+type ValidRuneTest struct {
+ r rune
+ ok bool
+}
+
+var validrunetests = []ValidRuneTest{
+ {0, true},
+ {'e', true},
+ {'é', true},
+ {'☺', true},
+ {RuneError, true},
+ {MaxRune, true},
+ {0xD7FF, true},
+ {0xD800, false},
+ {0xDFFF, false},
+ {0xE000, true},
+ {MaxRune + 1, false},
+ {-1, false},
+}
+
+func TestValidRune(t *testing.T) {
+ for _, tt := range validrunetests {
+ if ok := ValidRune(tt.r); ok != tt.ok {
+ t.Errorf("ValidRune(%#U) = %t, want %t", tt.r, ok, tt.ok)
+ }
+ }
+}
+
+func BenchmarkRuneCountTenASCIIChars(b *testing.B) {
+ for i := 0; i < b.N; i++ {
+ RuneCountInString("0123456789")
+ }
+}
+
+func BenchmarkRuneCountTenJapaneseChars(b *testing.B) {
+ for i := 0; i < b.N; i++ {
+ RuneCountInString("日本語日本語日本語日")
+ }
+}
+
+func BenchmarkEncodeASCIIRune(b *testing.B) {
+ buf := make([]byte, UTFMax)
+ for i := 0; i < b.N; i++ {
+ EncodeRune(buf, 'a')
+ }
+}
+
+func BenchmarkEncodeJapaneseRune(b *testing.B) {
+ buf := make([]byte, UTFMax)
+ for i := 0; i < b.N; i++ {
+ EncodeRune(buf, '本')
+ }
+}
+
+func BenchmarkDecodeASCIIRune(b *testing.B) {
+ a := []byte{'a'}
+ for i := 0; i < b.N; i++ {
+ DecodeRune(a)
+ }
+}
+
+func BenchmarkDecodeJapaneseRune(b *testing.B) {
+ nihon := []byte("本")
+ for i := 0; i < b.N; i++ {
+ DecodeRune(nihon)
+ }
+}