aboutsummaryrefslogtreecommitdiff
path: root/src/unicode
diff options
context:
space:
mode:
authorMarcel van Lohuizen <mpvl@golang.org>2019-03-27 13:27:01 +0100
committerMarcel van Lohuizen <mpvl@golang.org>2019-04-24 15:22:37 +0000
commite308d41fe51b691c4bb52edc0865ca49f938a0f9 (patch)
tree6b1a84e671f7324cb0a91dae3c8f0a55a5cb54e5 /src/unicode
parent80e7832733fd245181af3394077f2df21303a4aa (diff)
downloadgo-e308d41fe51b691c4bb52edc0865ca49f938a0f9.tar.gz
go-e308d41fe51b691c4bb52edc0865ca49f938a0f9.zip
unicode: update table using new generator in x/text
The changes in Unicode 11 exposes a bug in maketables.go. We update the Unicode 10 tables using a new generator to minimize the changes upgrading to Unicode 11. This change switches over the generation from core to that in x/text. To properly update the tables one needs to run the generate in x/text anyway, so this makes that official. The RangeTable generator in x/text also generates slightly compacter tables. Updates golang/go#27945 See CL 154443 Change-Id: I6c59e082d5b8cd9e9332a32d8971061228581d66 Reviewed-on: https://go-review.googlesource.com/c/go/+/169617 Run-TryBot: Marcel van Lohuizen <mpvl@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Diffstat (limited to 'src/unicode')
-rw-r--r--src/unicode/letter.go3
-rw-r--r--src/unicode/maketables.go1452
-rw-r--r--src/unicode/tables.go1218
3 files changed, 507 insertions, 2166 deletions
diff --git a/src/unicode/letter.go b/src/unicode/letter.go
index 8be9a7b7c9..a57566f0a5 100644
--- a/src/unicode/letter.go
+++ b/src/unicode/letter.go
@@ -6,9 +6,6 @@
// 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.
diff --git a/src/unicode/maketables.go b/src/unicode/maketables.go
deleted file mode 100644
index a1f15869ea..0000000000
--- a/src/unicode/maketables.go
+++ /dev/null
@@ -1,1452 +0,0 @@
-// 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()
-}
-
-func defaultVersion() string {
- if v := os.Getenv("UNICODE_VERSION"); v != "" {
- return v
- }
- return unicode.Version
-}
-
-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/"+defaultVersion()+"/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 https://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 https://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 > 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.
-
-// Code generated by maketables; DO NOT EDIT.
-// To regenerate, run:
-// maketables --tables=%s --data=%s --casefolding=%s
-
-package unicode
-
-`
-
-var categoryMapping = map[string]string{
- "Lu": "Letter, uppercase",
- "Ll": "Letter, lowercase",
- "Lt": "Letter, titlecase",
- "Lm": "Letter, modifier",
- "Lo": "Letter, other",
- "Mn": "Mark, nonspacing",
- "Mc": "Mark, spacing combining",
- "Me": "Mark, enclosing",
- "Nd": "Number, decimal digit",
- "Nl": "Number, letter",
- "No": "Number, other",
- "Pc": "Punctuation, connector",
- "Pd": "Punctuation, dash",
- "Ps": "Punctuation, open",
- "Pe": "Punctuation, close",
- "Pi": "Punctuation, initial quote",
- "Pf": "Punctuation, final quote",
- "Po": "Punctuation, other",
- "Sm": "Symbol, math",
- "Sc": "Symbol, currency",
- "Sk": "Symbol, modifier",
- "So": "Symbol, other",
- "Zs": "Separator, space",
- "Zl": "Separator, line",
- "Zp": "Separator, paragraph",
- "Cc": "Other, control",
- "Cf": "Other, format",
- "Cs": "Other, surrogate",
- "Co": "Other, private use",
- "Cn": "Other, not assigned",
-}
-
-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 {
- desc, ok := categoryMapping[name]
- if ok {
- varDecl += fmt.Sprintf(
- "\t%s = _%s; // %s is the set of Unicode characters in category %s (%s).\n",
- name, name, name, name, desc)
- } else {
- 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)
- }
- }
- }
- }
-}
-
-var deprecatedAliases = map[string]string{
- "Sentence_Terminal": "STerm",
-}
-
-// 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)
- if alias, ok := deprecatedAliases[k]; ok {
- printf("\t%q: %s,\n", alias, k)
- }
- }
- print("}\n\n")
- }
-
- decl := make(sort.StringSlice, len(list)+len(deprecatedAliases))
- 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++
- if alias, ok := deprecatedAliases[name]; ok {
- decl[ndecl] = fmt.Sprintf(
- "\t%[1]s = _%[2]s;\t// %[1]s is an alias for %[2]s.\n",
- alias, 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")
-}
-
-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] or [upper, lower] 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
- }
- if len(orb) == 2 && chars[orb[1]].upperCase == orb[0] && chars[orb[0]].lowerCase == orb[1] {
- caseOrbit[i] = nil
- }
- }
-
- // Record orbit information in chars.
- for _, orb := range caseOrbit {
- if orb == nil {
- continue
- }
- sort.Slice(orb, func(i, j int) bool {
- return orb[i] < orb[j]
- })
- c := orb[len(orb)-1]
- for _, d := range orb {
- chars[c].caseOrbit = d
- c = d
- }
- }
-
- printAsciiFold()
- 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 {
- scr[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 printAsciiFold() {
- printf("var asciiFold = [MaxASCII + 1]uint16{\n")
- for i := rune(0); i <= unicode.MaxASCII; i++ {
- 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
- }
- }
- printf("\t0x%04X,\n", f)
- }
- printf("}\n\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/tables.go b/src/unicode/tables.go
index ce85b128ca..90bfa10064 100644
--- a/src/unicode/tables.go
+++ b/src/unicode/tables.go
@@ -1,10 +1,4 @@
-// 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.
-
-// Code generated by maketables; DO NOT EDIT.
-// To regenerate, run:
-// maketables --tables=all --data=http://www.unicode.org/Public/10.0.0/ucd/UnicodeData.txt --casefolding=http://www.unicode.org/Public/10.0.0/ucd/CaseFolding.txt
+// Code generated by running "go generate" in golang.org/x/text. DO NOT EDIT.
package unicode
@@ -885,8 +879,7 @@ var _Lm = &RangeTable{
{0xff9f, 0xff9f, 1},
},
R32: []Range32{
- {0x16b40, 0x16b40, 1},
- {0x16b41, 0x16b43, 1},
+ {0x16b40, 0x16b43, 1},
{0x16f93, 0x16f9f, 1},
{0x16fe0, 0x16fe1, 1},
},
@@ -2799,8 +2792,7 @@ var _Po = &RangeTable{
{0xff65, 0xff65, 1},
},
R32: []Range32{
- {0x10100, 0x10100, 1},
- {0x10101, 0x10102, 1},
+ {0x10100, 0x10102, 1},
{0x1039f, 0x103d0, 49},
{0x1056f, 0x10857, 744},
{0x1091f, 0x1093f, 32},
@@ -3114,8 +3106,7 @@ var _Sk = &RangeTable{
{0xffe3, 0xffe3, 1},
},
R32: []Range32{
- {0x1f3fb, 0x1f3fb, 1},
- {0x1f3fc, 0x1f3ff, 1},
+ {0x1f3fb, 0x1f3ff, 1},
},
LatinOffset: 3,
}
@@ -3281,8 +3272,7 @@ var _So = &RangeTable{
{0xfffd, 0xfffd, 1},
},
R32: []Range32{
- {0x10137, 0x10137, 1},
- {0x10138, 0x1013f, 1},
+ {0x10137, 0x1013f, 1},
{0x10179, 0x10189, 1},
{0x1018c, 0x1018e, 1},
{0x10190, 0x1019b, 1},
@@ -3429,10 +3419,6 @@ var (
Zs = _Zs // Zs is the set of Unicode characters in category Zs (Separator, space).
)
-// Generated by running
-// maketables --scripts=all --url=http://www.unicode.org/Public/10.0.0/ucd/
-// DO NOT EDIT
-
// Scripts is the set of Unicode script tables.
var Scripts = map[string]*RangeTable{
"Adlam": Adlam,
@@ -3608,9 +3594,8 @@ var _Arabic = &RangeTable{
{0x0600, 0x0604, 1},
{0x0606, 0x060b, 1},
{0x060d, 0x061a, 1},
- {0x061c, 0x061c, 1},
- {0x061e, 0x061e, 1},
- {0x0620, 0x063f, 1},
+ {0x061c, 0x0620, 2},
+ {0x0621, 0x063f, 1},
{0x0641, 0x064a, 1},
{0x0656, 0x066f, 1},
{0x0671, 0x06dc, 1},
@@ -3633,32 +3618,23 @@ var _Arabic = &RangeTable{
{0x1ee00, 0x1ee03, 1},
{0x1ee05, 0x1ee1f, 1},
{0x1ee21, 0x1ee22, 1},
- {0x1ee24, 0x1ee24, 1},
- {0x1ee27, 0x1ee27, 1},
+ {0x1ee24, 0x1ee27, 3},
{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},
+ {0x1ee39, 0x1ee3b, 2},
+ {0x1ee42, 0x1ee47, 5},
+ {0x1ee49, 0x1ee4d, 2},
+ {0x1ee4e, 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},
+ {0x1ee54, 0x1ee57, 3},
+ {0x1ee59, 0x1ee61, 2},
+ {0x1ee62, 0x1ee64, 2},
{0x1ee67, 0x1ee6a, 1},
{0x1ee6c, 0x1ee72, 1},
{0x1ee74, 0x1ee77, 1},
{0x1ee79, 0x1ee7c, 1},
- {0x1ee7e, 0x1ee7e, 1},
- {0x1ee80, 0x1ee89, 1},
+ {0x1ee7e, 0x1ee80, 2},
+ {0x1ee81, 0x1ee89, 1},
{0x1ee8b, 0x1ee9b, 1},
{0x1eea1, 0x1eea3, 1},
{0x1eea5, 0x1eea9, 1},
@@ -3672,8 +3648,8 @@ var _Armenian = &RangeTable{
{0x0531, 0x0556, 1},
{0x0559, 0x055f, 1},
{0x0561, 0x0587, 1},
- {0x058a, 0x058a, 1},
- {0x058d, 0x058f, 1},
+ {0x058a, 0x058d, 3},
+ {0x058e, 0x058f, 1},
{0xfb13, 0xfb17, 1},
},
}
@@ -3724,14 +3700,14 @@ var _Bengali = &RangeTable{
{0x098f, 0x0990, 1},
{0x0993, 0x09a8, 1},
{0x09aa, 0x09b0, 1},
- {0x09b2, 0x09b2, 1},
- {0x09b6, 0x09b9, 1},
+ {0x09b2, 0x09b6, 4},
+ {0x09b7, 0x09b9, 1},
{0x09bc, 0x09c4, 1},
{0x09c7, 0x09c8, 1},
{0x09cb, 0x09ce, 1},
- {0x09d7, 0x09d7, 1},
- {0x09dc, 0x09dd, 1},
- {0x09df, 0x09e3, 1},
+ {0x09d7, 0x09dc, 5},
+ {0x09dd, 0x09df, 2},
+ {0x09e0, 0x09e3, 1},
{0x09e6, 0x09fd, 1},
},
}
@@ -3836,34 +3812,26 @@ var _Common = &RangeTable{
{0x007b, 0x00a9, 1},
{0x00ab, 0x00b9, 1},
{0x00bb, 0x00bf, 1},
- {0x00d7, 0x00d7, 1},
- {0x00f7, 0x00f7, 1},
+ {0x00d7, 0x00f7, 32},
{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, 0x061b, 1},
- {0x061f, 0x061f, 1},
- {0x0640, 0x0640, 1},
- {0x06dd, 0x06dd, 1},
- {0x08e2, 0x08e2, 1},
+ {0x0374, 0x037e, 10},
+ {0x0385, 0x0387, 2},
+ {0x0589, 0x0605, 124},
+ {0x060c, 0x061b, 15},
+ {0x061f, 0x0640, 33},
+ {0x06dd, 0x08e2, 517},
{0x0964, 0x0965, 1},
- {0x0e3f, 0x0e3f, 1},
- {0x0fd5, 0x0fd8, 1},
- {0x10fb, 0x10fb, 1},
- {0x16eb, 0x16ed, 1},
+ {0x0e3f, 0x0fd5, 406},
+ {0x0fd6, 0x0fd8, 1},
+ {0x10fb, 0x16eb, 1520},
+ {0x16ec, 0x16ed, 1},
{0x1735, 0x1736, 1},
{0x1802, 0x1803, 1},
- {0x1805, 0x1805, 1},
- {0x1cd3, 0x1cd3, 1},
- {0x1ce1, 0x1ce1, 1},
- {0x1ce9, 0x1cec, 1},
+ {0x1805, 0x1cd3, 1230},
+ {0x1ce1, 0x1ce9, 8},
+ {0x1cea, 0x1cec, 1},
{0x1cee, 0x1cf3, 1},
{0x1cf5, 0x1cf7, 1},
{0x2000, 0x200b, 1},
@@ -3890,14 +3858,14 @@ var _Common = &RangeTable{
{0x2e00, 0x2e49, 1},
{0x2ff0, 0x2ffb, 1},
{0x3000, 0x3004, 1},
- {0x3006, 0x3006, 1},
- {0x3008, 0x3020, 1},
+ {0x3006, 0x3008, 2},
+ {0x3009, 0x3020, 1},
{0x3030, 0x3037, 1},
{0x303c, 0x303f, 1},
{0x309b, 0x309c, 1},
- {0x30a0, 0x30a0, 1},
- {0x30fb, 0x30fc, 1},
- {0x3190, 0x319f, 1},
+ {0x30a0, 0x30fb, 91},
+ {0x30fc, 0x3190, 148},
+ {0x3191, 0x319f, 1},
{0x31c0, 0x31e3, 1},
{0x3220, 0x325f, 1},
{0x327f, 0x32cf, 1},
@@ -3906,21 +3874,20 @@ var _Common = &RangeTable{
{0xa700, 0xa721, 1},
{0xa788, 0xa78a, 1},
{0xa830, 0xa839, 1},
- {0xa92e, 0xa92e, 1},
- {0xa9cf, 0xa9cf, 1},
- {0xab5b, 0xab5b, 1},
- {0xfd3e, 0xfd3f, 1},
- {0xfe10, 0xfe19, 1},
+ {0xa92e, 0xa9cf, 161},
+ {0xab5b, 0xfd3e, 20963},
+ {0xfd3f, 0xfe10, 209},
+ {0xfe11, 0xfe19, 1},
{0xfe30, 0xfe52, 1},
{0xfe54, 0xfe66, 1},
{0xfe68, 0xfe6b, 1},
- {0xfeff, 0xfeff, 1},
- {0xff01, 0xff20, 1},
+ {0xfeff, 0xff01, 2},
+ {0xff02, 0xff20, 1},
{0xff3b, 0xff40, 1},
{0xff5b, 0xff65, 1},
- {0xff70, 0xff70, 1},
- {0xff9e, 0xff9f, 1},
- {0xffe0, 0xffe6, 1},
+ {0xff70, 0xff9e, 46},
+ {0xff9f, 0xffe0, 65},
+ {0xffe1, 0xffe6, 1},
{0xffe8, 0xffee, 1},
{0xfff9, 0xfffd, 1},
},
@@ -3944,12 +3911,12 @@ var _Common = &RangeTable{
{0x1d400, 0x1d454, 1},
{0x1d456, 0x1d49c, 1},
{0x1d49e, 0x1d49f, 1},
- {0x1d4a2, 0x1d4a2, 1},
- {0x1d4a5, 0x1d4a6, 1},
- {0x1d4a9, 0x1d4ac, 1},
+ {0x1d4a2, 0x1d4a5, 3},
+ {0x1d4a6, 0x1d4a9, 3},
+ {0x1d4aa, 0x1d4ac, 1},
{0x1d4ae, 0x1d4b9, 1},
- {0x1d4bb, 0x1d4bb, 1},
- {0x1d4bd, 0x1d4c3, 1},
+ {0x1d4bb, 0x1d4bd, 2},
+ {0x1d4be, 0x1d4c3, 1},
{0x1d4c5, 0x1d505, 1},
{0x1d507, 0x1d50a, 1},
{0x1d50d, 0x1d514, 1},
@@ -3957,8 +3924,8 @@ var _Common = &RangeTable{
{0x1d51e, 0x1d539, 1},
{0x1d53b, 0x1d53e, 1},
{0x1d540, 0x1d544, 1},
- {0x1d546, 0x1d546, 1},
- {0x1d54a, 0x1d550, 1},
+ {0x1d546, 0x1d54a, 4},
+ {0x1d54b, 0x1d550, 1},
{0x1d552, 0x1d6a5, 1},
{0x1d6a8, 0x1d7cb, 1},
{0x1d7ce, 0x1d7ff, 1},
@@ -3993,12 +3960,12 @@ var _Common = &RangeTable{
{0x1f940, 0x1f94c, 1},
{0x1f950, 0x1f96b, 1},
{0x1f980, 0x1f997, 1},
- {0x1f9c0, 0x1f9c0, 1},
- {0x1f9d0, 0x1f9e6, 1},
- {0xe0001, 0xe0001, 1},
- {0xe0020, 0xe007f, 1},
+ {0x1f9c0, 0x1f9d0, 16},
+ {0x1f9d1, 0x1f9e6, 1},
+ {0xe0001, 0xe0020, 31},
+ {0xe0021, 0xe007f, 1},
},
- LatinOffset: 7,
+ LatinOffset: 6,
}
var _Coptic = &RangeTable{
@@ -4023,11 +3990,10 @@ var _Cypriot = &RangeTable{
R16: []Range16{},
R32: []Range32{
{0x10800, 0x10805, 1},
- {0x10808, 0x10808, 1},
- {0x1080a, 0x10835, 1},
+ {0x10808, 0x1080a, 2},
+ {0x1080b, 0x10835, 1},
{0x10837, 0x10838, 1},
- {0x1083c, 0x1083c, 1},
- {0x1083f, 0x1083f, 1},
+ {0x1083c, 0x1083f, 3},
},
}
@@ -4036,8 +4002,7 @@ var _Cyrillic = &RangeTable{
{0x0400, 0x0484, 1},
{0x0487, 0x052f, 1},
{0x1c80, 0x1c88, 1},
- {0x1d2b, 0x1d2b, 1},
- {0x1d78, 0x1d78, 1},
+ {0x1d2b, 0x1d78, 77},
{0x2de0, 0x2dff, 1},
{0xa640, 0xa69f, 1},
{0xfe2e, 0xfe2f, 1},
@@ -4090,15 +4055,15 @@ var _Ethiopic = &RangeTable{
{0x1200, 0x1248, 1},
{0x124a, 0x124d, 1},
{0x1250, 0x1256, 1},
- {0x1258, 0x1258, 1},
- {0x125a, 0x125d, 1},
+ {0x1258, 0x125a, 2},
+ {0x125b, 0x125d, 1},
{0x1260, 0x1288, 1},
{0x128a, 0x128d, 1},
{0x1290, 0x12b0, 1},
{0x12b2, 0x12b5, 1},
{0x12b8, 0x12be, 1},
- {0x12c0, 0x12c0, 1},
- {0x12c2, 0x12c5, 1},
+ {0x12c0, 0x12c2, 2},
+ {0x12c3, 0x12c5, 1},
{0x12c8, 0x12d6, 1},
{0x12d8, 0x1310, 1},
{0x1312, 0x1315, 1},
@@ -4125,13 +4090,11 @@ var _Ethiopic = &RangeTable{
var _Georgian = &RangeTable{
R16: []Range16{
{0x10a0, 0x10c5, 1},
- {0x10c7, 0x10c7, 1},
- {0x10cd, 0x10cd, 1},
+ {0x10c7, 0x10cd, 6},
{0x10d0, 0x10fa, 1},
{0x10fc, 0x10ff, 1},
{0x2d00, 0x2d25, 1},
- {0x2d27, 0x2d27, 1},
- {0x2d2d, 0x2d2d, 1},
+ {0x2d27, 0x2d2d, 6},
},
}
@@ -4169,8 +4132,7 @@ var _Grantha = &RangeTable{
{0x1133c, 0x11344, 1},
{0x11347, 0x11348, 1},
{0x1134b, 0x1134d, 1},
- {0x11350, 0x11350, 1},
- {0x11357, 0x11357, 1},
+ {0x11350, 0x11357, 7},
{0x1135d, 0x11363, 1},
{0x11366, 0x1136c, 1},
{0x11370, 0x11374, 1},
@@ -4182,27 +4144,24 @@ var _Greek = &RangeTable{
{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},
+ {0x037f, 0x0384, 5},
+ {0x0386, 0x0388, 2},
+ {0x0389, 0x038a, 1},
+ {0x038c, 0x038e, 2},
+ {0x038f, 0x03a1, 1},
{0x03a3, 0x03e1, 1},
{0x03f0, 0x03ff, 1},
{0x1d26, 0x1d2a, 1},
{0x1d5d, 0x1d61, 1},
{0x1d66, 0x1d6a, 1},
- {0x1dbf, 0x1dbf, 1},
- {0x1f00, 0x1f15, 1},
+ {0x1dbf, 0x1f00, 321},
+ {0x1f01, 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},
+ {0x1f59, 0x1f5f, 2},
+ {0x1f60, 0x1f7d, 1},
{0x1f80, 0x1fb4, 1},
{0x1fb6, 0x1fc4, 1},
{0x1fc6, 0x1fd3, 1},
@@ -4210,13 +4169,12 @@ var _Greek = &RangeTable{
{0x1fdd, 0x1fef, 1},
{0x1ff2, 0x1ff4, 1},
{0x1ff6, 0x1ffe, 1},
- {0x2126, 0x2126, 1},
- {0xab65, 0xab65, 1},
+ {0x2126, 0xab65, 35391},
},
R32: []Range32{
{0x10140, 0x1018e, 1},
- {0x101a0, 0x101a0, 1},
- {0x1d200, 0x1d245, 1},
+ {0x101a0, 0x1d200, 53344},
+ {0x1d201, 0x1d245, 1},
},
}
@@ -4232,8 +4190,8 @@ var _Gujarati = &RangeTable{
{0x0abc, 0x0ac5, 1},
{0x0ac7, 0x0ac9, 1},
{0x0acb, 0x0acd, 1},
- {0x0ad0, 0x0ad0, 1},
- {0x0ae0, 0x0ae3, 1},
+ {0x0ad0, 0x0ae0, 16},
+ {0x0ae1, 0x0ae3, 1},
{0x0ae6, 0x0af1, 1},
{0x0af9, 0x0aff, 1},
},
@@ -4249,14 +4207,14 @@ var _Gurmukhi = &RangeTable{
{0x0a32, 0x0a33, 1},
{0x0a35, 0x0a36, 1},
{0x0a38, 0x0a39, 1},
- {0x0a3c, 0x0a3c, 1},
- {0x0a3e, 0x0a42, 1},
+ {0x0a3c, 0x0a3e, 2},
+ {0x0a3f, 0x0a42, 1},
{0x0a47, 0x0a48, 1},
{0x0a4b, 0x0a4d, 1},
- {0x0a51, 0x0a51, 1},
- {0x0a59, 0x0a5c, 1},
- {0x0a5e, 0x0a5e, 1},
- {0x0a66, 0x0a75, 1},
+ {0x0a51, 0x0a59, 8},
+ {0x0a5a, 0x0a5c, 1},
+ {0x0a5e, 0x0a66, 8},
+ {0x0a67, 0x0a75, 1},
},
}
@@ -4265,8 +4223,7 @@ var _Han = &RangeTable{
{0x2e80, 0x2e99, 1},
{0x2e9b, 0x2ef3, 1},
{0x2f00, 0x2fd5, 1},
- {0x3005, 0x3005, 1},
- {0x3007, 0x3007, 1},
+ {0x3005, 0x3007, 2},
{0x3021, 0x3029, 1},
{0x3038, 0x303b, 1},
{0x3400, 0x4db5, 1},
@@ -4325,10 +4282,10 @@ var _Hebrew = &RangeTable{
{0x05f0, 0x05f4, 1},
{0xfb1d, 0xfb36, 1},
{0xfb38, 0xfb3c, 1},
- {0xfb3e, 0xfb3e, 1},
- {0xfb40, 0xfb41, 1},
- {0xfb43, 0xfb44, 1},
- {0xfb46, 0xfb4f, 1},
+ {0xfb3e, 0xfb40, 2},
+ {0xfb41, 0xfb43, 2},
+ {0xfb44, 0xfb46, 2},
+ {0xfb47, 0xfb4f, 1},
},
}
@@ -4356,14 +4313,13 @@ var _Inherited = &RangeTable{
{0x0300, 0x036f, 1},
{0x0485, 0x0486, 1},
{0x064b, 0x0655, 1},
- {0x0670, 0x0670, 1},
- {0x0951, 0x0952, 1},
- {0x1ab0, 0x1abe, 1},
+ {0x0670, 0x0951, 737},
+ {0x0952, 0x1ab0, 4446},
+ {0x1ab1, 0x1abe, 1},
{0x1cd0, 0x1cd2, 1},
{0x1cd4, 0x1ce0, 1},
{0x1ce2, 0x1ce8, 1},
- {0x1ced, 0x1ced, 1},
- {0x1cf4, 0x1cf4, 1},
+ {0x1ced, 0x1cf4, 7},
{0x1cf8, 0x1cf9, 1},
{0x1dc0, 0x1df9, 1},
{0x1dfb, 0x1dff, 1},
@@ -4375,8 +4331,7 @@ var _Inherited = &RangeTable{
{0xfe20, 0xfe2d, 1},
},
R32: []Range32{
- {0x101fd, 0x101fd, 1},
- {0x102e0, 0x102e0, 1},
+ {0x101fd, 0x102e0, 227},
{0x1d167, 0x1d169, 1},
{0x1d17b, 0x1d182, 1},
{0x1d185, 0x1d18b, 1},
@@ -4428,8 +4383,8 @@ var _Kannada = &RangeTable{
{0x0cc6, 0x0cc8, 1},
{0x0cca, 0x0ccd, 1},
{0x0cd5, 0x0cd6, 1},
- {0x0cde, 0x0cde, 1},
- {0x0ce0, 0x0ce3, 1},
+ {0x0cde, 0x0ce0, 2},
+ {0x0ce1, 0x0ce3, 1},
{0x0ce6, 0x0cef, 1},
{0x0cf1, 0x0cf2, 1},
},
@@ -4499,21 +4454,19 @@ var _Khudawadi = &RangeTable{
var _Lao = &RangeTable{
R16: []Range16{
{0x0e81, 0x0e82, 1},
- {0x0e84, 0x0e84, 1},
- {0x0e87, 0x0e88, 1},
- {0x0e8a, 0x0e8a, 1},
- {0x0e8d, 0x0e8d, 1},
- {0x0e94, 0x0e97, 1},
+ {0x0e84, 0x0e87, 3},
+ {0x0e88, 0x0e8a, 2},
+ {0x0e8d, 0x0e94, 7},
+ {0x0e95, 0x0e97, 1},
{0x0e99, 0x0e9f, 1},
{0x0ea1, 0x0ea3, 1},
- {0x0ea5, 0x0ea5, 1},
- {0x0ea7, 0x0ea7, 1},
+ {0x0ea5, 0x0ea7, 2},
{0x0eaa, 0x0eab, 1},
{0x0ead, 0x0eb9, 1},
{0x0ebb, 0x0ebd, 1},
{0x0ec0, 0x0ec4, 1},
- {0x0ec6, 0x0ec6, 1},
- {0x0ec8, 0x0ecd, 1},
+ {0x0ec6, 0x0ec8, 2},
+ {0x0ec9, 0x0ecd, 1},
{0x0ed0, 0x0ed9, 1},
{0x0edc, 0x0edf, 1},
},
@@ -4523,8 +4476,7 @@ var _Latin = &RangeTable{
R16: []Range16{
{0x0041, 0x005a, 1},
{0x0061, 0x007a, 1},
- {0x00aa, 0x00aa, 1},
- {0x00ba, 0x00ba, 1},
+ {0x00aa, 0x00ba, 16},
{0x00c0, 0x00d6, 1},
{0x00d8, 0x00f6, 1},
{0x00f8, 0x02b8, 1},
@@ -4535,12 +4487,10 @@ var _Latin = &RangeTable{
{0x1d6b, 0x1d77, 1},
{0x1d79, 0x1dbe, 1},
{0x1e00, 0x1eff, 1},
- {0x2071, 0x2071, 1},
- {0x207f, 0x207f, 1},
+ {0x2071, 0x207f, 14},
{0x2090, 0x209c, 1},
{0x212a, 0x212b, 1},
- {0x2132, 0x2132, 1},
- {0x214e, 0x214e, 1},
+ {0x2132, 0x214e, 28},
{0x2160, 0x2188, 1},
{0x2c60, 0x2c7f, 1},
{0xa722, 0xa787, 1},
@@ -4553,7 +4503,7 @@ var _Latin = &RangeTable{
{0xff21, 0xff3a, 1},
{0xff41, 0xff5a, 1},
},
- LatinOffset: 6,
+ LatinOffset: 5,
}
var _Lepcha = &RangeTable{
@@ -4569,8 +4519,8 @@ var _Limbu = &RangeTable{
{0x1900, 0x191e, 1},
{0x1920, 0x192b, 1},
{0x1930, 0x193b, 1},
- {0x1940, 0x1940, 1},
- {0x1944, 0x194f, 1},
+ {0x1940, 0x1944, 4},
+ {0x1945, 0x194f, 1},
},
}
@@ -4667,9 +4617,9 @@ var _Masaram_Gondi = &RangeTable{
{0x11d00, 0x11d06, 1},
{0x11d08, 0x11d09, 1},
{0x11d0b, 0x11d36, 1},
- {0x11d3a, 0x11d3a, 1},
- {0x11d3c, 0x11d3d, 1},
- {0x11d3f, 0x11d47, 1},
+ {0x11d3a, 0x11d3c, 2},
+ {0x11d3d, 0x11d3f, 2},
+ {0x11d40, 0x11d47, 1},
{0x11d50, 0x11d59, 1},
},
}
@@ -4726,8 +4676,8 @@ var _Modi = &RangeTable{
var _Mongolian = &RangeTable{
R16: []Range16{
{0x1800, 0x1801, 1},
- {0x1804, 0x1804, 1},
- {0x1806, 0x180e, 1},
+ {0x1804, 0x1806, 2},
+ {0x1807, 0x180e, 1},
{0x1810, 0x1819, 1},
{0x1820, 0x1877, 1},
{0x1880, 0x18aa, 1},
@@ -4750,8 +4700,8 @@ var _Multani = &RangeTable{
R16: []Range16{},
R32: []Range32{
{0x11280, 0x11286, 1},
- {0x11288, 0x11288, 1},
- {0x1128a, 0x1128d, 1},
+ {0x11288, 0x1128a, 2},
+ {0x1128b, 0x1128d, 1},
{0x1128f, 0x1129d, 1},
{0x1129f, 0x112a9, 1},
},
@@ -4786,8 +4736,7 @@ var _Newa = &RangeTable{
R16: []Range16{},
R32: []Range32{
{0x11400, 0x11459, 1},
- {0x1145b, 0x1145b, 1},
- {0x1145d, 0x1145d, 1},
+ {0x1145b, 0x1145d, 2},
},
}
@@ -4800,8 +4749,8 @@ var _Nko = &RangeTable{
var _Nushu = &RangeTable{
R16: []Range16{},
R32: []Range32{
- {0x16fe1, 0x16fe1, 1},
- {0x1b170, 0x1b2fb, 1},
+ {0x16fe1, 0x1b170, 16783},
+ {0x1b171, 0x1b2fb, 1},
},
}
@@ -5019,12 +4968,12 @@ var _Sinhala = &RangeTable{
{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},
+ {0x0dbd, 0x0dc0, 3},
+ {0x0dc1, 0x0dc6, 1},
+ {0x0dca, 0x0dcf, 5},
+ {0x0dd0, 0x0dd4, 1},
+ {0x0dd6, 0x0dd8, 2},
+ {0x0dd9, 0x0ddf, 1},
{0x0de6, 0x0def, 1},
{0x0df2, 0x0df4, 1},
},
@@ -5126,16 +5075,15 @@ var _Tamil = &RangeTable{
{0x0b8e, 0x0b90, 1},
{0x0b92, 0x0b95, 1},
{0x0b99, 0x0b9a, 1},
- {0x0b9c, 0x0b9c, 1},
- {0x0b9e, 0x0b9f, 1},
- {0x0ba3, 0x0ba4, 1},
- {0x0ba8, 0x0baa, 1},
+ {0x0b9c, 0x0b9e, 2},
+ {0x0b9f, 0x0ba3, 4},
+ {0x0ba4, 0x0ba8, 4},
+ {0x0ba9, 0x0baa, 1},
{0x0bae, 0x0bb9, 1},
{0x0bbe, 0x0bc2, 1},
{0x0bc6, 0x0bc8, 1},
{0x0bca, 0x0bcd, 1},
- {0x0bd0, 0x0bd0, 1},
- {0x0bd7, 0x0bd7, 1},
+ {0x0bd0, 0x0bd7, 7},
{0x0be6, 0x0bfa, 1},
},
}
@@ -5143,8 +5091,8 @@ var _Tamil = &RangeTable{
var _Tangut = &RangeTable{
R16: []Range16{},
R32: []Range32{
- {0x16fe0, 0x16fe0, 1},
- {0x17000, 0x187ec, 1},
+ {0x16fe0, 0x17000, 32},
+ {0x17001, 0x187ec, 1},
{0x18800, 0x18af2, 1},
},
}
@@ -5389,10 +5337,6 @@ var (
Zanabazar_Square = _Zanabazar_Square // Zanabazar_Square is the set of Unicode characters in script Zanabazar_Square.
)
-// Generated by running
-// maketables --props=all --url=http://www.unicode.org/Public/10.0.0/ucd/
-// DO NOT EDIT
-
// Properties is the set of Unicode property tables.
var Properties = map[string]*RangeTable{
"ASCII_Hex_Digit": ASCII_Hex_Digit,
@@ -5443,46 +5387,35 @@ var _ASCII_Hex_Digit = &RangeTable{
var _Bidi_Control = &RangeTable{
R16: []Range16{
- {0x061c, 0x061c, 1},
- {0x200e, 0x200f, 1},
- {0x202a, 0x202e, 1},
+ {0x061c, 0x200e, 6642},
+ {0x200f, 0x202a, 27},
+ {0x202b, 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},
+ {0x002d, 0x058a, 1373},
+ {0x05be, 0x1400, 3650},
+ {0x1806, 0x2010, 2058},
+ {0x2011, 0x2015, 1},
+ {0x2053, 0x207b, 40},
+ {0x208b, 0x2212, 391},
+ {0x2e17, 0x2e1a, 3},
{0x2e3a, 0x2e3b, 1},
- {0x2e40, 0x2e40, 1},
- {0x301c, 0x301c, 1},
- {0x3030, 0x3030, 1},
- {0x30a0, 0x30a0, 1},
+ {0x2e40, 0x301c, 476},
+ {0x3030, 0x30a0, 112},
{0xfe31, 0xfe32, 1},
- {0xfe58, 0xfe58, 1},
- {0xfe63, 0xfe63, 1},
+ {0xfe58, 0xfe63, 11},
{0xff0d, 0xff0d, 1},
},
- LatinOffset: 1,
}
var _Deprecated = &RangeTable{
R16: []Range16{
- {0x0149, 0x0149, 1},
- {0x0673, 0x0673, 1},
- {0x0f77, 0x0f77, 1},
- {0x0f79, 0x0f79, 1},
+ {0x0149, 0x0673, 1322},
+ {0x0f77, 0x0f79, 2},
{0x17a3, 0x17a4, 1},
{0x206a, 0x206f, 1},
{0x2329, 0x232a, 1},
@@ -5494,25 +5427,22 @@ var _Deprecated = &RangeTable{
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},
+ {0x005e, 0x0060, 2},
+ {0x00a8, 0x00af, 7},
+ {0x00b4, 0x00b7, 3},
+ {0x00b8, 0x02b0, 504},
+ {0x02b1, 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},
+ {0x037a, 0x0384, 10},
+ {0x0385, 0x0483, 254},
+ {0x0484, 0x0487, 1},
+ {0x0559, 0x0591, 56},
+ {0x0592, 0x05a1, 1},
{0x05a3, 0x05bd, 1},
- {0x05bf, 0x05bf, 1},
- {0x05c1, 0x05c2, 1},
- {0x05c4, 0x05c4, 1},
+ {0x05bf, 0x05c1, 2},
+ {0x05c2, 0x05c4, 2},
{0x064b, 0x0652, 1},
{0x0657, 0x0658, 1},
{0x06df, 0x06e0, 1},
@@ -5523,129 +5453,104 @@ var _Diacritic = &RangeTable{
{0x07eb, 0x07f5, 1},
{0x0818, 0x0819, 1},
{0x08e3, 0x08fe, 1},
- {0x093c, 0x093c, 1},
- {0x094d, 0x094d, 1},
+ {0x093c, 0x094d, 17},
{0x0951, 0x0954, 1},
- {0x0971, 0x0971, 1},
- {0x09bc, 0x09bc, 1},
- {0x09cd, 0x09cd, 1},
- {0x0a3c, 0x0a3c, 1},
- {0x0a4d, 0x0a4d, 1},
- {0x0abc, 0x0abc, 1},
- {0x0acd, 0x0acd, 1},
- {0x0afd, 0x0aff, 1},
- {0x0b3c, 0x0b3c, 1},
- {0x0b4d, 0x0b4d, 1},
- {0x0bcd, 0x0bcd, 1},
- {0x0c4d, 0x0c4d, 1},
- {0x0cbc, 0x0cbc, 1},
- {0x0ccd, 0x0ccd, 1},
+ {0x0971, 0x09bc, 75},
+ {0x09cd, 0x0a3c, 111},
+ {0x0a4d, 0x0abc, 111},
+ {0x0acd, 0x0afd, 48},
+ {0x0afe, 0x0aff, 1},
+ {0x0b3c, 0x0b4d, 17},
+ {0x0bcd, 0x0c4d, 128},
+ {0x0cbc, 0x0ccd, 17},
{0x0d3b, 0x0d3c, 1},
- {0x0d4d, 0x0d4d, 1},
- {0x0dca, 0x0dca, 1},
- {0x0e47, 0x0e4c, 1},
- {0x0e4e, 0x0e4e, 1},
- {0x0ec8, 0x0ecc, 1},
+ {0x0d4d, 0x0e47, 125},
+ {0x0e48, 0x0e4c, 1},
+ {0x0e4e, 0x0ec8, 122},
+ {0x0ec9, 0x0ecc, 1},
{0x0f18, 0x0f19, 1},
- {0x0f35, 0x0f35, 1},
- {0x0f37, 0x0f37, 1},
- {0x0f39, 0x0f39, 1},
+ {0x0f35, 0x0f39, 2},
{0x0f3e, 0x0f3f, 1},
{0x0f82, 0x0f84, 1},
{0x0f86, 0x0f87, 1},
- {0x0fc6, 0x0fc6, 1},
- {0x1037, 0x1037, 1},
+ {0x0fc6, 0x1037, 113},
{0x1039, 0x103a, 1},
{0x1087, 0x108d, 1},
- {0x108f, 0x108f, 1},
- {0x109a, 0x109b, 1},
- {0x17c9, 0x17d3, 1},
- {0x17dd, 0x17dd, 1},
- {0x1939, 0x193b, 1},
+ {0x108f, 0x109a, 11},
+ {0x109b, 0x17c9, 1838},
+ {0x17ca, 0x17d3, 1},
+ {0x17dd, 0x1939, 348},
+ {0x193a, 0x193b, 1},
{0x1a75, 0x1a7c, 1},
- {0x1a7f, 0x1a7f, 1},
- {0x1ab0, 0x1abd, 1},
- {0x1b34, 0x1b34, 1},
- {0x1b44, 0x1b44, 1},
+ {0x1a7f, 0x1ab0, 49},
+ {0x1ab1, 0x1abd, 1},
+ {0x1b34, 0x1b44, 16},
{0x1b6b, 0x1b73, 1},
{0x1baa, 0x1bab, 1},
{0x1c36, 0x1c37, 1},
{0x1c78, 0x1c7d, 1},
{0x1cd0, 0x1ce8, 1},
- {0x1ced, 0x1ced, 1},
- {0x1cf4, 0x1cf4, 1},
+ {0x1ced, 0x1cf4, 7},
{0x1cf7, 0x1cf9, 1},
{0x1d2c, 0x1d6a, 1},
{0x1dc4, 0x1dcf, 1},
{0x1df5, 0x1df9, 1},
{0x1dfd, 0x1dff, 1},
- {0x1fbd, 0x1fbd, 1},
- {0x1fbf, 0x1fc1, 1},
+ {0x1fbd, 0x1fbf, 2},
+ {0x1fc0, 0x1fc1, 1},
{0x1fcd, 0x1fcf, 1},
{0x1fdd, 0x1fdf, 1},
{0x1fed, 0x1fef, 1},
{0x1ffd, 0x1ffe, 1},
{0x2cef, 0x2cf1, 1},
- {0x2e2f, 0x2e2f, 1},
- {0x302a, 0x302f, 1},
+ {0x2e2f, 0x302a, 507},
+ {0x302b, 0x302f, 1},
{0x3099, 0x309c, 1},
- {0x30fc, 0x30fc, 1},
- {0xa66f, 0xa66f, 1},
+ {0x30fc, 0xa66f, 30067},
{0xa67c, 0xa67d, 1},
- {0xa67f, 0xa67f, 1},
- {0xa69c, 0xa69d, 1},
- {0xa6f0, 0xa6f1, 1},
- {0xa717, 0xa721, 1},
- {0xa788, 0xa788, 1},
- {0xa7f8, 0xa7f9, 1},
- {0xa8c4, 0xa8c4, 1},
+ {0xa67f, 0xa69c, 29},
+ {0xa69d, 0xa6f0, 83},
+ {0xa6f1, 0xa717, 38},
+ {0xa718, 0xa721, 1},
+ {0xa788, 0xa7f8, 112},
+ {0xa7f9, 0xa8c4, 203},
{0xa8e0, 0xa8f1, 1},
{0xa92b, 0xa92e, 1},
- {0xa953, 0xa953, 1},
- {0xa9b3, 0xa9b3, 1},
- {0xa9c0, 0xa9c0, 1},
- {0xa9e5, 0xa9e5, 1},
+ {0xa953, 0xa9b3, 96},
+ {0xa9c0, 0xa9e5, 37},
{0xaa7b, 0xaa7d, 1},
{0xaabf, 0xaac2, 1},
- {0xaaf6, 0xaaf6, 1},
- {0xab5b, 0xab5f, 1},
+ {0xaaf6, 0xab5b, 101},
+ {0xab5c, 0xab5f, 1},
{0xabec, 0xabed, 1},
- {0xfb1e, 0xfb1e, 1},
- {0xfe20, 0xfe2f, 1},
- {0xff3e, 0xff3e, 1},
- {0xff40, 0xff40, 1},
- {0xff70, 0xff70, 1},
- {0xff9e, 0xff9f, 1},
- {0xffe3, 0xffe3, 1},
+ {0xfb1e, 0xfe20, 770},
+ {0xfe21, 0xfe2f, 1},
+ {0xff3e, 0xff40, 2},
+ {0xff70, 0xff9e, 46},
+ {0xff9f, 0xffe3, 68},
},
R32: []Range32{
- {0x102e0, 0x102e0, 1},
- {0x10ae5, 0x10ae6, 1},
- {0x110b9, 0x110ba, 1},
- {0x11133, 0x11134, 1},
- {0x11173, 0x11173, 1},
- {0x111c0, 0x111c0, 1},
- {0x111ca, 0x111cc, 1},
+ {0x102e0, 0x10ae5, 2053},
+ {0x10ae6, 0x110b9, 1491},
+ {0x110ba, 0x11133, 121},
+ {0x11134, 0x11173, 63},
+ {0x111c0, 0x111ca, 10},
+ {0x111cb, 0x111cc, 1},
{0x11235, 0x11236, 1},
{0x112e9, 0x112ea, 1},
- {0x1133c, 0x1133c, 1},
- {0x1134d, 0x1134d, 1},
+ {0x1133c, 0x1134d, 17},
{0x11366, 0x1136c, 1},
{0x11370, 0x11374, 1},
- {0x11442, 0x11442, 1},
- {0x11446, 0x11446, 1},
+ {0x11442, 0x11446, 4},
{0x114c2, 0x114c3, 1},
{0x115bf, 0x115c0, 1},
- {0x1163f, 0x1163f, 1},
- {0x116b6, 0x116b7, 1},
- {0x1172b, 0x1172b, 1},
- {0x11a34, 0x11a34, 1},
- {0x11a47, 0x11a47, 1},
- {0x11a99, 0x11a99, 1},
- {0x11c3f, 0x11c3f, 1},
- {0x11d42, 0x11d42, 1},
- {0x11d44, 0x11d45, 1},
- {0x16af0, 0x16af4, 1},
+ {0x1163f, 0x116b6, 119},
+ {0x116b7, 0x1172b, 116},
+ {0x11a34, 0x11a47, 19},
+ {0x11a99, 0x11c3f, 422},
+ {0x11d42, 0x11d44, 2},
+ {0x11d45, 0x16af0, 19883},
+ {0x16af1, 0x16af4, 1},
{0x16f8f, 0x16f9f, 1},
{0x1d167, 0x1d169, 1},
{0x1d16d, 0x1d172, 1},
@@ -5656,44 +5561,35 @@ var _Diacritic = &RangeTable{
{0x1e944, 0x1e946, 1},
{0x1e948, 0x1e94a, 1},
},
- LatinOffset: 6,
+ LatinOffset: 3,
}
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},
+ {0x00b7, 0x02d0, 537},
+ {0x02d1, 0x0640, 879},
+ {0x07fa, 0x0e46, 1612},
+ {0x0ec6, 0x180a, 2372},
+ {0x1843, 0x1aa7, 612},
+ {0x1c36, 0x1c7b, 69},
+ {0x3005, 0x3031, 44},
+ {0x3032, 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},
+ {0xa015, 0xa60c, 1527},
+ {0xa9cf, 0xa9e6, 23},
+ {0xaa70, 0xaadd, 109},
{0xaaf3, 0xaaf4, 1},
{0xff70, 0xff70, 1},
},
R32: []Range32{
- {0x1135d, 0x1135d, 1},
- {0x115c6, 0x115c8, 1},
- {0x11a98, 0x11a98, 1},
- {0x16b42, 0x16b43, 1},
- {0x16fe0, 0x16fe1, 1},
- {0x1e944, 0x1e946, 1},
+ {0x1135d, 0x115c6, 617},
+ {0x115c7, 0x115c8, 1},
+ {0x11a98, 0x16b42, 20650},
+ {0x16b43, 0x16fe0, 1181},
+ {0x16fe1, 0x1e944, 31075},
+ {0x1e945, 0x1e946, 1},
},
- LatinOffset: 1,
}
var _Hex_Digit = &RangeTable{
@@ -5710,18 +5606,14 @@ var _Hex_Digit = &RangeTable{
var _Hyphen = &RangeTable{
R16: []Range16{
- {0x002d, 0x002d, 1},
- {0x00ad, 0x00ad, 1},
- {0x058a, 0x058a, 1},
- {0x1806, 0x1806, 1},
+ {0x002d, 0x00ad, 128},
+ {0x058a, 0x1806, 4732},
{0x2010, 0x2011, 1},
- {0x2e17, 0x2e17, 1},
- {0x30fb, 0x30fb, 1},
- {0xfe63, 0xfe63, 1},
- {0xff0d, 0xff0d, 1},
+ {0x2e17, 0x30fb, 740},
+ {0xfe63, 0xff0d, 170},
{0xff65, 0xff65, 1},
},
- LatinOffset: 2,
+ LatinOffset: 1,
}
var _IDS_Binary_Operator = &RangeTable{
@@ -5771,9 +5663,8 @@ var _Logical_Order_Exception = &RangeTable{
{0x0e40, 0x0e44, 1},
{0x0ec0, 0x0ec4, 1},
{0x19b5, 0x19b7, 1},
- {0x19ba, 0x19ba, 1},
- {0xaab5, 0xaab6, 1},
- {0xaab9, 0xaab9, 1},
+ {0x19ba, 0xaab5, 37115},
+ {0xaab6, 0xaab9, 3},
{0xaabb, 0xaabc, 1},
},
}
@@ -5805,21 +5696,19 @@ var _Noncharacter_Code_Point = &RangeTable{
var _Other_Alphabetic = &RangeTable{
R16: []Range16{
- {0x0345, 0x0345, 1},
- {0x05b0, 0x05bd, 1},
- {0x05bf, 0x05bf, 1},
- {0x05c1, 0x05c2, 1},
- {0x05c4, 0x05c5, 1},
- {0x05c7, 0x05c7, 1},
+ {0x0345, 0x05b0, 619},
+ {0x05b1, 0x05bd, 1},
+ {0x05bf, 0x05c1, 2},
+ {0x05c2, 0x05c4, 2},
+ {0x05c5, 0x05c7, 2},
{0x0610, 0x061a, 1},
{0x064b, 0x0657, 1},
{0x0659, 0x065f, 1},
- {0x0670, 0x0670, 1},
- {0x06d6, 0x06dc, 1},
+ {0x0670, 0x06d6, 102},
+ {0x06d7, 0x06dc, 1},
{0x06e1, 0x06e4, 1},
{0x06e7, 0x06e8, 1},
- {0x06ed, 0x06ed, 1},
- {0x0711, 0x0711, 1},
+ {0x06ed, 0x0711, 36},
{0x0730, 0x073f, 1},
{0x07a6, 0x07b0, 1},
{0x0816, 0x0817, 1},
@@ -5838,15 +5727,14 @@ var _Other_Alphabetic = &RangeTable{
{0x09be, 0x09c4, 1},
{0x09c7, 0x09c8, 1},
{0x09cb, 0x09cc, 1},
- {0x09d7, 0x09d7, 1},
- {0x09e2, 0x09e3, 1},
- {0x0a01, 0x0a03, 1},
+ {0x09d7, 0x09e2, 11},
+ {0x09e3, 0x0a01, 30},
+ {0x0a02, 0x0a03, 1},
{0x0a3e, 0x0a42, 1},
{0x0a47, 0x0a48, 1},
{0x0a4b, 0x0a4c, 1},
- {0x0a51, 0x0a51, 1},
- {0x0a70, 0x0a71, 1},
- {0x0a75, 0x0a75, 1},
+ {0x0a51, 0x0a70, 31},
+ {0x0a71, 0x0a75, 4},
{0x0a81, 0x0a83, 1},
{0x0abe, 0x0ac5, 1},
{0x0ac7, 0x0ac9, 1},
@@ -5859,12 +5747,12 @@ var _Other_Alphabetic = &RangeTable{
{0x0b4b, 0x0b4c, 1},
{0x0b56, 0x0b57, 1},
{0x0b62, 0x0b63, 1},
- {0x0b82, 0x0b82, 1},
- {0x0bbe, 0x0bc2, 1},
+ {0x0b82, 0x0bbe, 60},
+ {0x0bbf, 0x0bc2, 1},
{0x0bc6, 0x0bc8, 1},
{0x0bca, 0x0bcc, 1},
- {0x0bd7, 0x0bd7, 1},
- {0x0c00, 0x0c03, 1},
+ {0x0bd7, 0x0c00, 41},
+ {0x0c01, 0x0c03, 1},
{0x0c3e, 0x0c44, 1},
{0x0c46, 0x0c48, 1},
{0x0c4a, 0x0c4c, 1},
@@ -5880,42 +5768,41 @@ var _Other_Alphabetic = &RangeTable{
{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},
+ {0x0d57, 0x0d62, 11},
+ {0x0d63, 0x0d82, 31},
+ {0x0d83, 0x0dcf, 76},
+ {0x0dd0, 0x0dd4, 1},
+ {0x0dd6, 0x0dd8, 2},
+ {0x0dd9, 0x0ddf, 1},
{0x0df2, 0x0df3, 1},
- {0x0e31, 0x0e31, 1},
- {0x0e34, 0x0e3a, 1},
- {0x0e4d, 0x0e4d, 1},
- {0x0eb1, 0x0eb1, 1},
+ {0x0e31, 0x0e34, 3},
+ {0x0e35, 0x0e3a, 1},
+ {0x0e4d, 0x0eb1, 100},
{0x0eb4, 0x0eb9, 1},
{0x0ebb, 0x0ebc, 1},
- {0x0ecd, 0x0ecd, 1},
- {0x0f71, 0x0f81, 1},
+ {0x0ecd, 0x0f71, 164},
+ {0x0f72, 0x0f81, 1},
{0x0f8d, 0x0f97, 1},
{0x0f99, 0x0fbc, 1},
{0x102b, 0x1036, 1},
- {0x1038, 0x1038, 1},
- {0x103b, 0x103e, 1},
+ {0x1038, 0x103b, 3},
+ {0x103c, 0x103e, 1},
{0x1056, 0x1059, 1},
{0x105e, 0x1060, 1},
- {0x1062, 0x1062, 1},
- {0x1067, 0x1068, 1},
- {0x1071, 0x1074, 1},
+ {0x1062, 0x1067, 5},
+ {0x1068, 0x1071, 9},
+ {0x1072, 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},
+ {0x135f, 0x1712, 947},
+ {0x1713, 0x1732, 31},
+ {0x1733, 0x1752, 31},
+ {0x1753, 0x1772, 31},
+ {0x1773, 0x17b6, 67},
+ {0x17b7, 0x17c8, 1},
{0x1885, 0x1886, 1},
- {0x18a9, 0x18a9, 1},
- {0x1920, 0x192b, 1},
+ {0x18a9, 0x1920, 119},
+ {0x1921, 0x192b, 1},
{0x1930, 0x1938, 1},
{0x1a17, 0x1a1b, 1},
{0x1a55, 0x1a5e, 1},
@@ -5936,21 +5823,20 @@ var _Other_Alphabetic = &RangeTable{
{0xa823, 0xa827, 1},
{0xa880, 0xa881, 1},
{0xa8b4, 0xa8c3, 1},
- {0xa8c5, 0xa8c5, 1},
- {0xa926, 0xa92a, 1},
+ {0xa8c5, 0xa926, 97},
+ {0xa927, 0xa92a, 1},
{0xa947, 0xa952, 1},
{0xa980, 0xa983, 1},
{0xa9b4, 0xa9bf, 1},
{0xaa29, 0xaa36, 1},
- {0xaa43, 0xaa43, 1},
- {0xaa4c, 0xaa4d, 1},
- {0xaab0, 0xaab0, 1},
+ {0xaa43, 0xaa4c, 9},
+ {0xaa4d, 0xaab0, 99},
{0xaab2, 0xaab4, 1},
{0xaab7, 0xaab8, 1},
- {0xaabe, 0xaabe, 1},
- {0xaaeb, 0xaaef, 1},
- {0xaaf5, 0xaaf5, 1},
- {0xabe3, 0xabea, 1},
+ {0xaabe, 0xaaeb, 45},
+ {0xaaec, 0xaaef, 1},
+ {0xaaf5, 0xabe3, 238},
+ {0xabe4, 0xabea, 1},
{0xfb1e, 0xfb1e, 1},
},
R32: []Range32{
@@ -5960,31 +5846,30 @@ var _Other_Alphabetic = &RangeTable{
{0x10a0c, 0x10a0f, 1},
{0x11000, 0x11002, 1},
{0x11038, 0x11045, 1},
- {0x11082, 0x11082, 1},
- {0x110b0, 0x110b8, 1},
+ {0x11082, 0x110b0, 46},
+ {0x110b1, 0x110b8, 1},
{0x11100, 0x11102, 1},
{0x11127, 0x11132, 1},
{0x11180, 0x11182, 1},
{0x111b3, 0x111bf, 1},
{0x1122c, 0x11234, 1},
- {0x11237, 0x11237, 1},
- {0x1123e, 0x1123e, 1},
+ {0x11237, 0x1123e, 7},
{0x112df, 0x112e8, 1},
{0x11300, 0x11303, 1},
{0x1133e, 0x11344, 1},
{0x11347, 0x11348, 1},
{0x1134b, 0x1134c, 1},
- {0x11357, 0x11357, 1},
- {0x11362, 0x11363, 1},
- {0x11435, 0x11441, 1},
+ {0x11357, 0x11362, 11},
+ {0x11363, 0x11435, 210},
+ {0x11436, 0x11441, 1},
{0x11443, 0x11445, 1},
{0x114b0, 0x114c1, 1},
{0x115af, 0x115b5, 1},
{0x115b8, 0x115be, 1},
{0x115dc, 0x115dd, 1},
{0x11630, 0x1163e, 1},
- {0x11640, 0x11640, 1},
- {0x116ab, 0x116b5, 1},
+ {0x11640, 0x116ab, 107},
+ {0x116ac, 0x116b5, 1},
{0x1171d, 0x1172a, 1},
{0x11a01, 0x11a0a, 1},
{0x11a35, 0x11a39, 1},
@@ -5996,21 +5881,20 @@ var _Other_Alphabetic = &RangeTable{
{0x11c92, 0x11ca7, 1},
{0x11ca9, 0x11cb6, 1},
{0x11d31, 0x11d36, 1},
- {0x11d3a, 0x11d3a, 1},
- {0x11d3c, 0x11d3d, 1},
- {0x11d3f, 0x11d41, 1},
- {0x11d43, 0x11d43, 1},
- {0x11d47, 0x11d47, 1},
+ {0x11d3a, 0x11d3c, 2},
+ {0x11d3d, 0x11d3f, 2},
+ {0x11d40, 0x11d41, 1},
+ {0x11d43, 0x11d47, 4},
{0x16b30, 0x16b36, 1},
{0x16f51, 0x16f7e, 1},
- {0x1bc9e, 0x1bc9e, 1},
- {0x1e000, 0x1e006, 1},
+ {0x1bc9e, 0x1e000, 9058},
+ {0x1e001, 0x1e006, 1},
{0x1e008, 0x1e018, 1},
{0x1e01b, 0x1e021, 1},
{0x1e023, 0x1e024, 1},
{0x1e026, 0x1e02a, 1},
- {0x1e947, 0x1e947, 1},
- {0x1f130, 0x1f149, 1},
+ {0x1e947, 0x1f130, 2025},
+ {0x1f131, 0x1f149, 1},
{0x1f150, 0x1f169, 1},
{0x1f170, 0x1f189, 1},
},
@@ -6018,17 +5902,15 @@ var _Other_Alphabetic = &RangeTable{
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},
+ {0x034f, 0x115f, 3600},
+ {0x1160, 0x17b4, 1620},
+ {0x17b5, 0x2065, 2224},
+ {0x3164, 0xffa0, 52796},
{0xfff0, 0xfff8, 1},
},
R32: []Range32{
- {0xe0000, 0xe0000, 1},
- {0xe0002, 0xe001f, 1},
+ {0xe0000, 0xe0002, 2},
+ {0xe0003, 0xe001f, 1},
{0xe0080, 0xe00ff, 1},
{0xe01f0, 0xe0fff, 1},
},
@@ -6036,29 +5918,20 @@ var _Other_Default_Ignorable_Code_Point = &RangeTable{
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, 0x200c, 1},
+ {0x09be, 0x09d7, 25},
+ {0x0b3e, 0x0b57, 25},
+ {0x0bbe, 0x0bd7, 25},
+ {0x0cc2, 0x0cd5, 19},
+ {0x0cd6, 0x0d3e, 104},
+ {0x0d57, 0x0dcf, 120},
+ {0x0ddf, 0x200c, 4653},
{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},
+ {0x1133e, 0x11357, 25},
+ {0x114b0, 0x114bd, 13},
+ {0x115af, 0x1d165, 48054},
{0x1d16e, 0x1d172, 1},
{0xe0020, 0xe007f, 1},
},
@@ -6066,75 +5939,67 @@ var _Other_Grapheme_Extend = &RangeTable{
var _Other_ID_Continue = &RangeTable{
R16: []Range16{
- {0x00b7, 0x00b7, 1},
- {0x0387, 0x0387, 1},
+ {0x00b7, 0x0387, 720},
{0x1369, 0x1371, 1},
{0x19da, 0x19da, 1},
},
- LatinOffset: 1,
}
var _Other_ID_Start = &RangeTable{
R16: []Range16{
{0x1885, 0x1886, 1},
- {0x2118, 0x2118, 1},
- {0x212e, 0x212e, 1},
+ {0x2118, 0x212e, 22},
{0x309b, 0x309c, 1},
},
}
var _Other_Lowercase = &RangeTable{
R16: []Range16{
- {0x00aa, 0x00aa, 1},
- {0x00ba, 0x00ba, 1},
+ {0x00aa, 0x00ba, 16},
{0x02b0, 0x02b8, 1},
{0x02c0, 0x02c1, 1},
{0x02e0, 0x02e4, 1},
- {0x0345, 0x0345, 1},
- {0x037a, 0x037a, 1},
+ {0x0345, 0x037a, 53},
{0x1d2c, 0x1d6a, 1},
- {0x1d78, 0x1d78, 1},
- {0x1d9b, 0x1dbf, 1},
- {0x2071, 0x2071, 1},
- {0x207f, 0x207f, 1},
+ {0x1d78, 0x1d9b, 35},
+ {0x1d9c, 0x1dbf, 1},
+ {0x2071, 0x207f, 14},
{0x2090, 0x209c, 1},
{0x2170, 0x217f, 1},
{0x24d0, 0x24e9, 1},
{0x2c7c, 0x2c7d, 1},
{0xa69c, 0xa69d, 1},
- {0xa770, 0xa770, 1},
- {0xa7f8, 0xa7f9, 1},
- {0xab5c, 0xab5f, 1},
+ {0xa770, 0xa7f8, 136},
+ {0xa7f9, 0xab5c, 867},
+ {0xab5d, 0xab5f, 1},
},
- LatinOffset: 2,
+ LatinOffset: 1,
}
var _Other_Math = &RangeTable{
R16: []Range16{
- {0x005e, 0x005e, 1},
- {0x03d0, 0x03d2, 1},
- {0x03d5, 0x03d5, 1},
- {0x03f0, 0x03f1, 1},
- {0x03f4, 0x03f5, 1},
- {0x2016, 0x2016, 1},
+ {0x005e, 0x03d0, 882},
+ {0x03d1, 0x03d2, 1},
+ {0x03d5, 0x03f0, 27},
+ {0x03f1, 0x03f4, 3},
+ {0x03f5, 0x2016, 7201},
{0x2032, 0x2034, 1},
- {0x2040, 0x2040, 1},
- {0x2061, 0x2064, 1},
+ {0x2040, 0x2061, 33},
+ {0x2062, 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},
+ {0x20e1, 0x20e5, 4},
+ {0x20e6, 0x20eb, 5},
+ {0x20ec, 0x20ef, 1},
+ {0x2102, 0x2107, 5},
{0x210a, 0x2113, 1},
- {0x2115, 0x2115, 1},
- {0x2119, 0x211d, 1},
- {0x2124, 0x2124, 1},
- {0x2128, 0x2129, 1},
- {0x212c, 0x212d, 1},
- {0x212f, 0x2131, 1},
+ {0x2115, 0x2119, 4},
+ {0x211a, 0x211d, 1},
+ {0x2124, 0x2128, 4},
+ {0x2129, 0x212c, 3},
+ {0x212d, 0x212f, 2},
+ {0x2130, 0x2131, 1},
{0x2133, 0x2138, 1},
{0x213c, 0x213f, 1},
{0x2145, 0x2149, 1},
@@ -6142,33 +6007,30 @@ var _Other_Math = &RangeTable{
{0x219c, 0x219f, 1},
{0x21a1, 0x21a2, 1},
{0x21a4, 0x21a5, 1},
- {0x21a7, 0x21a7, 1},
- {0x21a9, 0x21ad, 1},
+ {0x21a7, 0x21a9, 2},
+ {0x21aa, 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},
+ {0x21d3, 0x21d5, 2},
+ {0x21d6, 0x21db, 1},
+ {0x21dd, 0x21e4, 7},
+ {0x21e5, 0x2308, 291},
+ {0x2309, 0x230b, 1},
{0x23b4, 0x23b5, 1},
- {0x23b7, 0x23b7, 1},
- {0x23d0, 0x23d0, 1},
- {0x23e2, 0x23e2, 1},
- {0x25a0, 0x25a1, 1},
- {0x25ae, 0x25b6, 1},
+ {0x23b7, 0x23d0, 25},
+ {0x23e2, 0x25a0, 446},
+ {0x25a1, 0x25ae, 13},
+ {0x25af, 0x25b6, 1},
{0x25bc, 0x25c0, 1},
{0x25c6, 0x25c7, 1},
{0x25ca, 0x25cb, 1},
{0x25cf, 0x25d3, 1},
- {0x25e2, 0x25e2, 1},
- {0x25e4, 0x25e4, 1},
+ {0x25e2, 0x25e4, 2},
{0x25e7, 0x25ec, 1},
{0x2605, 0x2606, 1},
- {0x2640, 0x2640, 1},
- {0x2642, 0x2642, 1},
+ {0x2640, 0x2642, 2},
{0x2660, 0x2663, 1},
{0x266d, 0x266e, 1},
{0x27c5, 0x27c6, 1},
@@ -6176,22 +6038,20 @@ var _Other_Math = &RangeTable{
{0x2983, 0x2998, 1},
{0x29d8, 0x29db, 1},
{0x29fc, 0x29fd, 1},
- {0xfe61, 0xfe61, 1},
- {0xfe63, 0xfe63, 1},
- {0xfe68, 0xfe68, 1},
- {0xff3c, 0xff3c, 1},
+ {0xfe61, 0xfe63, 2},
+ {0xfe68, 0xff3c, 212},
{0xff3e, 0xff3e, 1},
},
R32: []Range32{
{0x1d400, 0x1d454, 1},
{0x1d456, 0x1d49c, 1},
{0x1d49e, 0x1d49f, 1},
- {0x1d4a2, 0x1d4a2, 1},
- {0x1d4a5, 0x1d4a6, 1},
- {0x1d4a9, 0x1d4ac, 1},
+ {0x1d4a2, 0x1d4a5, 3},
+ {0x1d4a6, 0x1d4a9, 3},
+ {0x1d4aa, 0x1d4ac, 1},
{0x1d4ae, 0x1d4b9, 1},
- {0x1d4bb, 0x1d4bb, 1},
- {0x1d4bd, 0x1d4c3, 1},
+ {0x1d4bb, 0x1d4bd, 2},
+ {0x1d4be, 0x1d4c3, 1},
{0x1d4c5, 0x1d505, 1},
{0x1d507, 0x1d50a, 1},
{0x1d50d, 0x1d514, 1},
@@ -6199,8 +6059,8 @@ var _Other_Math = &RangeTable{
{0x1d51e, 0x1d539, 1},
{0x1d53b, 0x1d53e, 1},
{0x1d540, 0x1d544, 1},
- {0x1d546, 0x1d546, 1},
- {0x1d54a, 0x1d550, 1},
+ {0x1d546, 0x1d54a, 4},
+ {0x1d54b, 0x1d550, 1},
{0x1d552, 0x1d6a5, 1},
{0x1d6a8, 0x1d6c0, 1},
{0x1d6c2, 0x1d6da, 1},
@@ -6217,38 +6077,28 @@ var _Other_Math = &RangeTable{
{0x1ee00, 0x1ee03, 1},
{0x1ee05, 0x1ee1f, 1},
{0x1ee21, 0x1ee22, 1},
- {0x1ee24, 0x1ee24, 1},
- {0x1ee27, 0x1ee27, 1},
+ {0x1ee24, 0x1ee27, 3},
{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},
+ {0x1ee39, 0x1ee3b, 2},
+ {0x1ee42, 0x1ee47, 5},
+ {0x1ee49, 0x1ee4d, 2},
+ {0x1ee4e, 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},
+ {0x1ee54, 0x1ee57, 3},
+ {0x1ee59, 0x1ee61, 2},
+ {0x1ee62, 0x1ee64, 2},
{0x1ee67, 0x1ee6a, 1},
{0x1ee6c, 0x1ee72, 1},
{0x1ee74, 0x1ee77, 1},
{0x1ee79, 0x1ee7c, 1},
- {0x1ee7e, 0x1ee7e, 1},
- {0x1ee80, 0x1ee89, 1},
+ {0x1ee7e, 0x1ee80, 2},
+ {0x1ee81, 0x1ee89, 1},
{0x1ee8b, 0x1ee9b, 1},
{0x1eea1, 0x1eea3, 1},
{0x1eea5, 0x1eea9, 1},
{0x1eeab, 0x1eebb, 1},
},
- LatinOffset: 1,
}
var _Other_Uppercase = &RangeTable{
@@ -6268,19 +6118,15 @@ var _Pattern_Syntax = &RangeTable{
{0x0021, 0x002f, 1},
{0x003a, 0x0040, 1},
{0x005b, 0x005e, 1},
- {0x0060, 0x0060, 1},
- {0x007b, 0x007e, 1},
+ {0x0060, 0x007b, 27},
+ {0x007c, 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},
+ {0x00a9, 0x00ab, 2},
+ {0x00ac, 0x00b0, 2},
+ {0x00b1, 0x00bb, 5},
+ {0x00bf, 0x00d7, 24},
+ {0x00f7, 0x2010, 7961},
+ {0x2011, 0x2027, 1},
{0x2030, 0x203e, 1},
{0x2041, 0x2053, 1},
{0x2055, 0x205e, 1},
@@ -6290,29 +6136,27 @@ var _Pattern_Syntax = &RangeTable{
{0x2e00, 0x2e7f, 1},
{0x3001, 0x3003, 1},
{0x3008, 0x3020, 1},
- {0x3030, 0x3030, 1},
- {0xfd3e, 0xfd3f, 1},
- {0xfe45, 0xfe46, 1},
+ {0x3030, 0xfd3e, 52494},
+ {0xfd3f, 0xfe45, 262},
+ {0xfe46, 0xfe46, 1},
},
- LatinOffset: 15,
+ LatinOffset: 10,
}
var _Pattern_White_Space = &RangeTable{
R16: []Range16{
{0x0009, 0x000d, 1},
- {0x0020, 0x0020, 1},
- {0x0085, 0x0085, 1},
+ {0x0020, 0x0085, 101},
{0x200e, 0x200f, 1},
{0x2028, 0x2029, 1},
},
- LatinOffset: 3,
+ LatinOffset: 2,
}
var _Prepended_Concatenation_Mark = &RangeTable{
R16: []Range16{
{0x0600, 0x0605, 1},
- {0x06dd, 0x06dd, 1},
- {0x070f, 0x070f, 1},
+ {0x06dd, 0x070f, 50},
{0x08e2, 0x08e2, 1},
},
R32: []Range32{
@@ -6322,21 +6166,18 @@ var _Prepended_Concatenation_Mark = &RangeTable{
var _Quotation_Mark = &RangeTable{
R16: []Range16{
- {0x0022, 0x0022, 1},
- {0x0027, 0x0027, 1},
- {0x00ab, 0x00ab, 1},
- {0x00bb, 0x00bb, 1},
+ {0x0022, 0x0027, 5},
+ {0x00ab, 0x00bb, 16},
{0x2018, 0x201f, 1},
{0x2039, 0x203a, 1},
- {0x2e42, 0x2e42, 1},
- {0x300c, 0x300f, 1},
+ {0x2e42, 0x300c, 458},
+ {0x300d, 0x300f, 1},
{0x301d, 0x301f, 1},
{0xfe41, 0xfe44, 1},
- {0xff02, 0xff02, 1},
- {0xff07, 0xff07, 1},
+ {0xff02, 0xff07, 5},
{0xff62, 0xff63, 1},
},
- LatinOffset: 4,
+ LatinOffset: 2,
}
var _Radical = &RangeTable{
@@ -6356,50 +6197,39 @@ var _Regional_Indicator = &RangeTable{
var _Sentence_Terminal = &RangeTable{
R16: []Range16{
- {0x0021, 0x0021, 1},
- {0x002e, 0x002e, 1},
- {0x003f, 0x003f, 1},
- {0x0589, 0x0589, 1},
- {0x061f, 0x061f, 1},
- {0x06d4, 0x06d4, 1},
+ {0x0021, 0x002e, 13},
+ {0x003f, 0x0589, 1354},
+ {0x061f, 0x06d4, 181},
{0x0700, 0x0702, 1},
- {0x07f9, 0x07f9, 1},
- {0x0964, 0x0965, 1},
- {0x104a, 0x104b, 1},
- {0x1362, 0x1362, 1},
+ {0x07f9, 0x0964, 363},
+ {0x0965, 0x104a, 1765},
+ {0x104b, 0x1362, 791},
{0x1367, 0x1368, 1},
- {0x166e, 0x166e, 1},
- {0x1735, 0x1736, 1},
- {0x1803, 0x1803, 1},
- {0x1809, 0x1809, 1},
- {0x1944, 0x1945, 1},
- {0x1aa8, 0x1aab, 1},
+ {0x166e, 0x1735, 199},
+ {0x1736, 0x1803, 205},
+ {0x1809, 0x1944, 315},
+ {0x1945, 0x1aa8, 355},
+ {0x1aa9, 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},
+ {0x2e2e, 0x2e3c, 14},
+ {0x3002, 0xa4ff, 29949},
{0xa60e, 0xa60f, 1},
- {0xa6f3, 0xa6f3, 1},
- {0xa6f7, 0xa6f7, 1},
+ {0xa6f3, 0xa6f7, 4},
{0xa876, 0xa877, 1},
{0xa8ce, 0xa8cf, 1},
- {0xa92f, 0xa92f, 1},
- {0xa9c8, 0xa9c9, 1},
- {0xaa5d, 0xaa5f, 1},
+ {0xa92f, 0xa9c8, 153},
+ {0xa9c9, 0xaa5d, 148},
+ {0xaa5e, 0xaa5f, 1},
{0xaaf0, 0xaaf1, 1},
- {0xabeb, 0xabeb, 1},
- {0xfe52, 0xfe52, 1},
+ {0xabeb, 0xfe52, 21095},
{0xfe56, 0xfe57, 1},
- {0xff01, 0xff01, 1},
- {0xff0e, 0xff0e, 1},
- {0xff1f, 0xff1f, 1},
- {0xff61, 0xff61, 1},
+ {0xff01, 0xff0e, 13},
+ {0xff1f, 0xff61, 66},
},
R32: []Range32{
{0x10a56, 0x10a57, 1},
@@ -6407,11 +6237,10 @@ var _Sentence_Terminal = &RangeTable{
{0x110be, 0x110c1, 1},
{0x11141, 0x11143, 1},
{0x111c5, 0x111c6, 1},
- {0x111cd, 0x111cd, 1},
- {0x111de, 0x111df, 1},
- {0x11238, 0x11239, 1},
- {0x1123b, 0x1123c, 1},
- {0x112a9, 0x112a9, 1},
+ {0x111cd, 0x111de, 17},
+ {0x111df, 0x11238, 89},
+ {0x11239, 0x1123b, 2},
+ {0x1123c, 0x112a9, 109},
{0x1144b, 0x1144c, 1},
{0x115c2, 0x115c3, 1},
{0x115c9, 0x115d7, 1},
@@ -6421,35 +6250,25 @@ var _Sentence_Terminal = &RangeTable{
{0x11a9b, 0x11a9c, 1},
{0x11c41, 0x11c42, 1},
{0x16a6e, 0x16a6f, 1},
- {0x16af5, 0x16af5, 1},
- {0x16b37, 0x16b38, 1},
- {0x16b44, 0x16b44, 1},
- {0x1bc9f, 0x1bc9f, 1},
- {0x1da88, 0x1da88, 1},
+ {0x16af5, 0x16b37, 66},
+ {0x16b38, 0x16b44, 12},
+ {0x1bc9f, 0x1da88, 7657},
},
- LatinOffset: 3,
+ LatinOffset: 1,
}
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},
+ {0x012f, 0x0249, 282},
+ {0x0268, 0x029d, 53},
+ {0x02b2, 0x03f3, 321},
+ {0x0456, 0x0458, 2},
+ {0x1d62, 0x1d96, 52},
+ {0x1da4, 0x1da8, 4},
+ {0x1e2d, 0x1ecb, 158},
+ {0x2071, 0x2148, 215},
+ {0x2149, 0x2c7c, 2867},
},
R32: []Range32{
{0x1d422, 0x1d423, 1},
@@ -6471,27 +6290,20 @@ var _Soft_Dotted = &RangeTable{
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},
+ {0x0021, 0x002c, 11},
+ {0x002e, 0x003a, 12},
+ {0x003b, 0x003f, 4},
+ {0x037e, 0x0387, 9},
+ {0x0589, 0x05c3, 58},
+ {0x060c, 0x061b, 15},
+ {0x061f, 0x06d4, 181},
{0x0700, 0x070a, 1},
- {0x070c, 0x070c, 1},
- {0x07f8, 0x07f9, 1},
- {0x0830, 0x083e, 1},
- {0x085e, 0x085e, 1},
- {0x0964, 0x0965, 1},
- {0x0e5a, 0x0e5b, 1},
- {0x0f08, 0x0f08, 1},
+ {0x070c, 0x07f8, 236},
+ {0x07f9, 0x0830, 55},
+ {0x0831, 0x083e, 1},
+ {0x085e, 0x0964, 262},
+ {0x0965, 0x0e5a, 1269},
+ {0x0e5b, 0x0f08, 173},
{0x0f0d, 0x0f12, 1},
{0x104a, 0x104b, 1},
{0x1361, 0x1368, 1},
@@ -6499,8 +6311,8 @@ var _Terminal_Punctuation = &RangeTable{
{0x16eb, 0x16ed, 1},
{0x1735, 0x1736, 1},
{0x17d4, 0x17d6, 1},
- {0x17da, 0x17da, 1},
- {0x1802, 0x1805, 1},
+ {0x17da, 0x1802, 40},
+ {0x1803, 0x1805, 1},
{0x1808, 0x1809, 1},
{0x1944, 0x1945, 1},
{0x1aa8, 0x1aab, 1},
@@ -6510,36 +6322,29 @@ var _Terminal_Punctuation = &RangeTable{
{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},
+ {0x2e2e, 0x2e3c, 14},
+ {0x2e41, 0x3001, 448},
+ {0x3002, 0xa4fe, 29948},
+ {0xa4ff, 0xa60d, 270},
+ {0xa60e, 0xa60f, 1},
{0xa6f3, 0xa6f7, 1},
{0xa876, 0xa877, 1},
{0xa8ce, 0xa8cf, 1},
- {0xa92f, 0xa92f, 1},
- {0xa9c7, 0xa9c9, 1},
+ {0xa92f, 0xa9c7, 152},
+ {0xa9c8, 0xa9c9, 1},
{0xaa5d, 0xaa5f, 1},
- {0xaadf, 0xaadf, 1},
- {0xaaf0, 0xaaf1, 1},
- {0xabeb, 0xabeb, 1},
+ {0xaadf, 0xaaf0, 17},
+ {0xaaf1, 0xabeb, 250},
{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},
+ {0xff01, 0xff0c, 11},
+ {0xff0e, 0xff1a, 12},
+ {0xff1b, 0xff1f, 4},
+ {0xff61, 0xff64, 3},
},
R32: []Range32{
- {0x1039f, 0x1039f, 1},
- {0x103d0, 0x103d0, 1},
- {0x10857, 0x10857, 1},
- {0x1091f, 0x1091f, 1},
+ {0x1039f, 0x103d0, 49},
+ {0x10857, 0x1091f, 200},
{0x10a56, 0x10a57, 1},
{0x10af0, 0x10af5, 1},
{0x10b3a, 0x10b3f, 1},
@@ -6548,13 +6353,13 @@ var _Terminal_Punctuation = &RangeTable{
{0x110be, 0x110c1, 1},
{0x11141, 0x11143, 1},
{0x111c5, 0x111c6, 1},
- {0x111cd, 0x111cd, 1},
- {0x111de, 0x111df, 1},
- {0x11238, 0x1123c, 1},
- {0x112a9, 0x112a9, 1},
- {0x1144b, 0x1144d, 1},
- {0x1145b, 0x1145b, 1},
- {0x115c2, 0x115c5, 1},
+ {0x111cd, 0x111de, 17},
+ {0x111df, 0x11238, 89},
+ {0x11239, 0x1123c, 1},
+ {0x112a9, 0x1144b, 418},
+ {0x1144c, 0x1144d, 1},
+ {0x1145b, 0x115c2, 359},
+ {0x115c3, 0x115c5, 1},
{0x115c9, 0x115d7, 1},
{0x11641, 0x11642, 1},
{0x1173c, 0x1173e, 1},
@@ -6562,16 +6367,15 @@ var _Terminal_Punctuation = &RangeTable{
{0x11a9b, 0x11a9c, 1},
{0x11aa1, 0x11aa2, 1},
{0x11c41, 0x11c43, 1},
- {0x11c71, 0x11c71, 1},
- {0x12470, 0x12474, 1},
+ {0x11c71, 0x12470, 2047},
+ {0x12471, 0x12474, 1},
{0x16a6e, 0x16a6f, 1},
- {0x16af5, 0x16af5, 1},
- {0x16b37, 0x16b39, 1},
- {0x16b44, 0x16b44, 1},
- {0x1bc9f, 0x1bc9f, 1},
+ {0x16af5, 0x16b37, 66},
+ {0x16b38, 0x16b39, 1},
+ {0x16b44, 0x1bc9f, 20827},
{0x1da87, 0x1da8a, 1},
},
- LatinOffset: 5,
+ LatinOffset: 3,
}
var _Unified_Ideograph = &RangeTable{
@@ -6579,12 +6383,11 @@ var _Unified_Ideograph = &RangeTable{
{0x3400, 0x4db5, 1},
{0x4e00, 0x9fea, 1},
{0xfa0e, 0xfa0f, 1},
- {0xfa11, 0xfa11, 1},
- {0xfa13, 0xfa14, 1},
- {0xfa1f, 0xfa1f, 1},
- {0xfa21, 0xfa21, 1},
- {0xfa23, 0xfa24, 1},
- {0xfa27, 0xfa29, 1},
+ {0xfa11, 0xfa13, 2},
+ {0xfa14, 0xfa1f, 11},
+ {0xfa21, 0xfa23, 2},
+ {0xfa24, 0xfa27, 3},
+ {0xfa28, 0xfa29, 1},
},
R32: []Range32{
{0x20000, 0x2a6d6, 1},
@@ -6608,17 +6411,14 @@ var _Variation_Selector = &RangeTable{
var _White_Space = &RangeTable{
R16: []Range16{
{0x0009, 0x000d, 1},
- {0x0020, 0x0020, 1},
- {0x0085, 0x0085, 1},
- {0x00a0, 0x00a0, 1},
- {0x1680, 0x1680, 1},
+ {0x0020, 0x0085, 101},
+ {0x00a0, 0x1680, 5600},
{0x2000, 0x200a, 1},
{0x2028, 0x2029, 1},
- {0x202f, 0x202f, 1},
- {0x205f, 0x205f, 1},
+ {0x202f, 0x205f, 48},
{0x3000, 0x3000, 1},
},
- LatinOffset: 4,
+ LatinOffset: 2,
}
// These variables have type *RangeTable.
@@ -6660,10 +6460,6 @@ var (
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/10.0.0/ucd/UnicodeData.txt --casefolding=http://www.unicode.org/Public/10.0.0/ucd/CaseFolding.txt
-// DO NOT EDIT
-
// CaseRanges is the table describing case mappings for all letters with
// non-self mappings.
var CaseRanges = _CaseRanges
@@ -7760,7 +7556,7 @@ var foldInherited = &RangeTable{
},
}
-// Range entries: 3587 16-bit, 1554 32-bit, 5141 total.
-// Range bytes: 21522 16-bit, 18648 32-bit, 40170 total.
+// Range entries: 3461 16-bit, 1511 32-bit, 4972 total.
+// Range bytes: 20766 16-bit, 18132 32-bit, 38898 total.
// Fold orbit bytes: 88 pairs, 352 bytes