mirrors/ceph-csi
1
0
Fork 0
mirror of https://github.com/ceph/ceph-csi.git synced 2025-06-13 10:33:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

Migrate from dep to go module

Browse Source 
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
This commit is contained in:
Madhu Rajanna
2020-02-17 17:45:57 +05:30
committed by mergify[bot]
parent a9174dd953
commit d5a0606c33
642 changed files with 54160 additions and 147015 deletions
Download Patch File Download Diff File Expand all files Collapse all files

702
vendor/golang.org/x/text/collate/build/builder.go generated vendored
View File

@ -1,702 +0,0 @@
// 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 build // import "golang.org/x/text/collate/build"
import (
"fmt"
"io"
"log"
"sort"
"strings"
"unicode/utf8"
"golang.org/x/text/internal/colltab"
"golang.org/x/text/language"
"golang.org/x/text/unicode/norm"
)
// TODO: optimizations:
// - expandElem is currently 20K. By putting unique colElems in a separate
// table and having a byte array of indexes into this table, we can reduce
// the total size to about 7K. By also factoring out the length bytes, we
// can reduce this to about 6K.
// - trie valueBlocks are currently 100K. There are a lot of sparse blocks
// and many consecutive values with the same stride. This can be further
// compacted.
// - Compress secondary weights into 8 bits.
// - Some LDML specs specify a context element. Currently we simply concatenate
// those. Context can be implemented using the contraction trie. If Builder
// could analyze and detect when using a context makes sense, there is no
// need to expose this construct in the API.
// A Builder builds a root collation table. The user must specify the
// collation elements for each entry. A common use will be to base the weights
// on those specified in the allkeys* file as provided by the UCA or CLDR.
type Builder struct {
index *trieBuilder
root ordering
locale []*Tailoring
t *table
err error
built bool
minNonVar int // lowest primary recorded for a variable
varTop int // highest primary recorded for a non-variable
// indexes used for reusing expansions and contractions
expIndex map[string]int // positions of expansions keyed by their string representation
ctHandle map[string]ctHandle // contraction handles keyed by a concatenation of the suffixes
ctElem map[string]int // contraction elements keyed by their string representation
}
// A Tailoring builds a collation table based on another collation table.
// The table is defined by specifying tailorings to the underlying table.
// See https://unicode.org/reports/tr35/ for an overview of tailoring
// collation tables. The CLDR contains pre-defined tailorings for a variety
// of languages (See https://www.unicode.org/Public/cldr/<version>/core.zip.)
type Tailoring struct {
id string
builder *Builder
index *ordering
anchor *entry
before bool
}
// NewBuilder returns a new Builder.
func NewBuilder() *Builder {
return &Builder{
index: newTrieBuilder(),
root: makeRootOrdering(),
expIndex: make(map[string]int),
ctHandle: make(map[string]ctHandle),
ctElem: make(map[string]int),
}
}
// Tailoring returns a Tailoring for the given locale. One should
// have completed all calls to Add before calling Tailoring.
func (b *Builder) Tailoring(loc language.Tag) *Tailoring {
t := &Tailoring{
id: loc.String(),
builder: b,
index: b.root.clone(),
}
t.index.id = t.id
b.locale = append(b.locale, t)
return t
}
// Add adds an entry to the collation element table, mapping
// a slice of runes to a sequence of collation elements.
// A collation element is specified as list of weights: []int{primary, secondary, ...}.
// The entries are typically obtained from a collation element table
// as defined in https://www.unicode.org/reports/tr10/#Data_Table_Format.
// Note that the collation elements specified by colelems are only used
// as a guide. The actual weights generated by Builder may differ.
// The argument variables is a list of indices into colelems that should contain
// a value for each colelem that is a variable. (See the reference above.)
func (b *Builder) Add(runes []rune, colelems [][]int, variables []int) error {
str := string(runes)
elems := make([]rawCE, len(colelems))
for i, ce := range colelems {
if len(ce) == 0 {
break
}
elems[i] = makeRawCE(ce, 0)
if len(ce) == 1 {
elems[i].w[1] = defaultSecondary
}
if len(ce) <= 2 {
elems[i].w[2] = defaultTertiary
}
if len(ce) <= 3 {
elems[i].w[3] = ce[0]
}
}
for i, ce := range elems {
p := ce.w[0]
isvar := false
for _, j := range variables {
if i == j {
isvar = true
}
}
if isvar {
if p >= b.minNonVar && b.minNonVar > 0 {
return fmt.Errorf("primary value %X of variable is larger than the smallest non-variable %X", p, b.minNonVar)
}
if p > b.varTop {
b.varTop = p
}
} else if p > 1 { // 1 is a special primary value reserved for FFFE
if p <= b.varTop {
return fmt.Errorf("primary value %X of non-variable is smaller than the highest variable %X", p, b.varTop)
}
if b.minNonVar == 0 || p < b.minNonVar {
b.minNonVar = p
}
}
}
elems, err := convertLargeWeights(elems)
if err != nil {
return err
}
cccs := []uint8{}
nfd := norm.NFD.String(str)
for i := range nfd {
cccs = append(cccs, norm.NFD.PropertiesString(nfd[i:]).CCC())
}
if len(cccs) < len(elems) {
if len(cccs) > 2 {
return fmt.Errorf("number of decomposed characters should be greater or equal to the number of collation elements for len(colelems) > 3 (%d < %d)", len(cccs), len(elems))
}
p := len(elems) - 1
for ; p > 0 && elems[p].w[0] == 0; p-- {
elems[p].ccc = cccs[len(cccs)-1]
}
for ; p >= 0; p-- {
elems[p].ccc = cccs[0]
}
} else {
for i := range elems {
elems[i].ccc = cccs[i]
}
}
// doNorm in collate.go assumes that the following conditions hold.
if len(elems) > 1 && len(cccs) > 1 && cccs[0] != 0 && cccs[0] != cccs[len(cccs)-1] {
return fmt.Errorf("incompatible CCC values for expansion %X (%d)", runes, cccs)
}
b.root.newEntry(str, elems)
return nil
}
func (t *Tailoring) setAnchor(anchor string) error {
anchor = norm.NFC.String(anchor)
a := t.index.find(anchor)
if a == nil {
a = t.index.newEntry(anchor, nil)
a.implicit = true
a.modified = true
for _, r := range []rune(anchor) {
e := t.index.find(string(r))
e.lock = true
}
}
t.anchor = a
return nil
}
// SetAnchor sets the point after which elements passed in subsequent calls to
// Insert will be inserted. It is equivalent to the reset directive in an LDML
// specification. See Insert for an example.
// SetAnchor supports the following logical reset positions:
// <first_tertiary_ignorable/>, <last_teriary_ignorable/>, <first_primary_ignorable/>,
// and <last_non_ignorable/>.
func (t *Tailoring) SetAnchor(anchor string) error {
if err := t.setAnchor(anchor); err != nil {
return err
}
t.before = false
return nil
}
// SetAnchorBefore is similar to SetAnchor, except that subsequent calls to
// Insert will insert entries before the anchor.
func (t *Tailoring) SetAnchorBefore(anchor string) error {
if err := t.setAnchor(anchor); err != nil {
return err
}
t.before = true
return nil
}
// Insert sets the ordering of str relative to the entry set by the previous
// call to SetAnchor or Insert. The argument extend corresponds
// to the extend elements as defined in LDML. A non-empty value for extend
// will cause the collation elements corresponding to extend to be appended
// to the collation elements generated for the entry added by Insert.
// This has the same net effect as sorting str after the string anchor+extend.
// See https://www.unicode.org/reports/tr10/#Tailoring_Example for details
// on parametric tailoring and https://unicode.org/reports/tr35/#Collation_Elements
// for full details on LDML.
//
// Examples: create a tailoring for Swedish, where "ä" is ordered after "z"
// at the primary sorting level:
// t := b.Tailoring("se")
// t.SetAnchor("z")
// t.Insert(colltab.Primary, "ä", "")
// Order "ü" after "ue" at the secondary sorting level:
// t.SetAnchor("ue")
// t.Insert(colltab.Secondary, "ü","")
// or
// t.SetAnchor("u")
// t.Insert(colltab.Secondary, "ü", "e")
// Order "q" afer "ab" at the secondary level and "Q" after "q"
// at the tertiary level:
// t.SetAnchor("ab")
// t.Insert(colltab.Secondary, "q", "")
// t.Insert(colltab.Tertiary, "Q", "")
// Order "b" before "a":
// t.SetAnchorBefore("a")
// t.Insert(colltab.Primary, "b", "")
// Order "0" after the last primary ignorable:
// t.SetAnchor("<last_primary_ignorable/>")
// t.Insert(colltab.Primary, "0", "")
func (t *Tailoring) Insert(level colltab.Level, str, extend string) error {
if t.anchor == nil {
return fmt.Errorf("%s:Insert: no anchor point set for tailoring of %s", t.id, str)
}
str = norm.NFC.String(str)
e := t.index.find(str)
if e == nil {
e = t.index.newEntry(str, nil)
} else if e.logical != noAnchor {
return fmt.Errorf("%s:Insert: cannot reinsert logical reset position %q", t.id, e.str)
}
if e.lock {
return fmt.Errorf("%s:Insert: cannot reinsert element %q", t.id, e.str)
}
a := t.anchor
// Find the first element after the anchor which differs at a level smaller or
// equal to the given level. Then insert at this position.
// See https://unicode.org/reports/tr35/#Collation_Elements, Section 5.14.5 for details.
e.before = t.before
if t.before {
t.before = false
if a.prev == nil {
a.insertBefore(e)
} else {
for a = a.prev; a.level > level; a = a.prev {
}
a.insertAfter(e)
}
e.level = level
} else {
for ; a.level > level; a = a.next {
}
e.level = a.level
if a != e {
a.insertAfter(e)
a.level = level
} else {
// We don't set a to prev itself. This has the effect of the entry
// getting new collation elements that are an increment of itself.
// This is intentional.
a.prev.level = level
}
}
e.extend = norm.NFD.String(extend)
e.exclude = false
e.modified = true
e.elems = nil
t.anchor = e
return nil
}
func (o *ordering) getWeight(e *entry) []rawCE {
if len(e.elems) == 0 && e.logical == noAnchor {
if e.implicit {
for _, r := range e.runes {
e.elems = append(e.elems, o.getWeight(o.find(string(r)))...)
}
} else if e.before {
count := [colltab.Identity + 1]int{}
a := e
for ; a.elems == nil && !a.implicit; a = a.next {
count[a.level]++
}
e.elems = []rawCE{makeRawCE(a.elems[0].w, a.elems[0].ccc)}
for i := colltab.Primary; i < colltab.Quaternary; i++ {
if count[i] != 0 {
e.elems[0].w[i] -= count[i]
break
}
}
if e.prev != nil {
o.verifyWeights(e.prev, e, e.prev.level)
}
} else {
prev := e.prev
e.elems = nextWeight(prev.level, o.getWeight(prev))
o.verifyWeights(e, e.next, e.level)
}
}
return e.elems
}
func (o *ordering) addExtension(e *entry) {
if ex := o.find(e.extend); ex != nil {
e.elems = append(e.elems, ex.elems...)
} else {
for _, r := range []rune(e.extend) {
e.elems = append(e.elems, o.find(string(r)).elems...)
}
}
e.extend = ""
}
func (o *ordering) verifyWeights(a, b *entry, level colltab.Level) error {
if level == colltab.Identity || b == nil || b.elems == nil || a.elems == nil {
return nil
}
for i := colltab.Primary; i < level; i++ {
if a.elems[0].w[i] < b.elems[0].w[i] {
return nil
}
}
if a.elems[0].w[level] >= b.elems[0].w[level] {
err := fmt.Errorf("%s:overflow: collation elements of %q (%X) overflows those of %q (%X) at level %d (%X >= %X)", o.id, a.str, a.runes, b.str, b.runes, level, a.elems, b.elems)
log.Println(err)
// TODO: return the error instead, or better, fix the conflicting entry by making room.
}
return nil
}
func (b *Builder) error(e error) {
if e != nil {
b.err = e
}
}
func (b *Builder) errorID(locale string, e error) {
if e != nil {
b.err = fmt.Errorf("%s:%v", locale, e)
}
}
// patchNorm ensures that NFC and NFD counterparts are consistent.
func (o *ordering) patchNorm() {
// Insert the NFD counterparts, if necessary.
for _, e := range o.ordered {
nfd := norm.NFD.String(e.str)
if nfd != e.str {
if e0 := o.find(nfd); e0 != nil && !e0.modified {
e0.elems = e.elems
} else if e.modified && !equalCEArrays(o.genColElems(nfd), e.elems) {
e := o.newEntry(nfd, e.elems)
e.modified = true
}
}
}
// Update unchanged composed forms if one of their parts changed.
for _, e := range o.ordered {
nfd := norm.NFD.String(e.str)
if e.modified || nfd == e.str {
continue
}
if e0 := o.find(nfd); e0 != nil {
e.elems = e0.elems
} else {
e.elems = o.genColElems(nfd)
if norm.NFD.LastBoundary([]byte(nfd)) == 0 {
r := []rune(nfd)
head := string(r[0])
tail := ""
for i := 1; i < len(r); i++ {
s := norm.NFC.String(head + string(r[i]))
if e0 := o.find(s); e0 != nil && e0.modified {
head = s
} else {
tail += string(r[i])
}
}
e.elems = append(o.genColElems(head), o.genColElems(tail)...)
}
}
}
// Exclude entries for which the individual runes generate the same collation elements.
for _, e := range o.ordered {
if len(e.runes) > 1 && equalCEArrays(o.genColElems(e.str), e.elems) {
e.exclude = true
}
}
}
func (b *Builder) buildOrdering(o *ordering) {
for _, e := range o.ordered {
o.getWeight(e)
}
for _, e := range o.ordered {
o.addExtension(e)
}
o.patchNorm()
o.sort()
simplify(o)
b.processExpansions(o) // requires simplify
b.processContractions(o) // requires simplify
t := newNode()
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
if !e.skip() {
ce, err := e.encode()
b.errorID(o.id, err)
t.insert(e.runes[0], ce)
}
}
o.handle = b.index.addTrie(t)
}
func (b *Builder) build() (*table, error) {
if b.built {
return b.t, b.err
}
b.built = true
b.t = &table{
Table: colltab.Table{
MaxContractLen: utf8.UTFMax,
VariableTop: uint32(b.varTop),
},
}
b.buildOrdering(&b.root)
b.t.root = b.root.handle
for _, t := range b.locale {
b.buildOrdering(t.index)
if b.err != nil {
break
}
}
i, err := b.index.generate()
b.t.trie = *i
b.t.Index = colltab.Trie{
Index: i.index,
Values: i.values,
Index0: i.index[blockSize*b.t.root.lookupStart:],
Values0: i.values[blockSize*b.t.root.valueStart:],
}
b.error(err)
return b.t, b.err
}
// Build builds the root Collator.
func (b *Builder) Build() (colltab.Weighter, error) {
table, err := b.build()
if err != nil {
return nil, err
}
return table, nil
}
// Build builds a Collator for Tailoring t.
func (t *Tailoring) Build() (colltab.Weighter, error) {
// TODO: implement.
return nil, nil
}
// Print prints the tables for b and all its Tailorings as a Go file
// that can be included in the Collate package.
func (b *Builder) Print(w io.Writer) (n int, err error) {
p := func(nn int, e error) {
n += nn
if err == nil {
err = e
}
}
t, err := b.build()
if err != nil {
return 0, err
}
p(fmt.Fprintf(w, `var availableLocales = "und`))
for _, loc := range b.locale {
if loc.id != "und" {
p(fmt.Fprintf(w, ",%s", loc.id))
}
}
p(fmt.Fprint(w, "\"\n\n"))
p(fmt.Fprintf(w, "const varTop = 0x%x\n\n", b.varTop))
p(fmt.Fprintln(w, "var locales = [...]tableIndex{"))
for _, loc := range b.locale {
if loc.id == "und" {
p(t.fprintIndex(w, loc.index.handle, loc.id))
}
}
for _, loc := range b.locale {
if loc.id != "und" {
p(t.fprintIndex(w, loc.index.handle, loc.id))
}
}
p(fmt.Fprint(w, "}\n\n"))
n, _, err = t.fprint(w, "main")
return
}
// reproducibleFromNFKD checks whether the given expansion could be generated
// from an NFKD expansion.
func reproducibleFromNFKD(e *entry, exp, nfkd []rawCE) bool {
// Length must be equal.
if len(exp) != len(nfkd) {
return false
}
for i, ce := range exp {
// Primary and secondary values should be equal.
if ce.w[0] != nfkd[i].w[0] || ce.w[1] != nfkd[i].w[1] {
return false
}
// Tertiary values should be equal to maxTertiary for third element onwards.
// TODO: there seem to be a lot of cases in CLDR (e.g. ㏭ in zh.xml) that can
// simply be dropped. Try this out by dropping the following code.
if i >= 2 && ce.w[2] != maxTertiary {
return false
}
if _, err := makeCE(ce); err != nil {
// Simply return false. The error will be caught elsewhere.
return false
}
}
return true
}
func simplify(o *ordering) {
// Runes that are a starter of a contraction should not be removed.
// (To date, there is only Kannada character 0CCA.)
keep := make(map[rune]bool)
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
if len(e.runes) > 1 {
keep[e.runes[0]] = true
}
}
// Tag entries for which the runes NFKD decompose to identical values.
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
s := e.str
nfkd := norm.NFKD.String(s)
nfd := norm.NFD.String(s)
if e.decompose || len(e.runes) > 1 || len(e.elems) == 1 || keep[e.runes[0]] || nfkd == nfd {
continue
}
if reproducibleFromNFKD(e, e.elems, o.genColElems(nfkd)) {
e.decompose = true
}
}
}
// appendExpansion converts the given collation sequence to
// collation elements and adds them to the expansion table.
// It returns an index to the expansion table.
func (b *Builder) appendExpansion(e *entry) int {
t := b.t
i := len(t.ExpandElem)
ce := uint32(len(e.elems))
t.ExpandElem = append(t.ExpandElem, ce)
for _, w := range e.elems {
ce, err := makeCE(w)
if err != nil {
b.error(err)
return -1
}
t.ExpandElem = append(t.ExpandElem, ce)
}
return i
}
// processExpansions extracts data necessary to generate
// the extraction tables.
func (b *Builder) processExpansions(o *ordering) {
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
if !e.expansion() {
continue
}
key := fmt.Sprintf("%v", e.elems)
i, ok := b.expIndex[key]
if !ok {
i = b.appendExpansion(e)
b.expIndex[key] = i
}
e.expansionIndex = i
}
}
func (b *Builder) processContractions(o *ordering) {
// Collate contractions per starter rune.
starters := []rune{}
cm := make(map[rune][]*entry)
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
if e.contraction() {
if len(e.str) > b.t.MaxContractLen {
b.t.MaxContractLen = len(e.str)
}
r := e.runes[0]
if _, ok := cm[r]; !ok {
starters = append(starters, r)
}
cm[r] = append(cm[r], e)
}
}
// Add entries of single runes that are at a start of a contraction.
for e := o.front(); e != nil; e, _ = e.nextIndexed() {
if !e.contraction() {
r := e.runes[0]
if _, ok := cm[r]; ok {
cm[r] = append(cm[r], e)
}
}
}
// Build the tries for the contractions.
t := b.t
for _, r := range starters {
l := cm[r]
// Compute suffix strings. There are 31 different contraction suffix
// sets for 715 contractions and 82 contraction starter runes as of
// version 6.0.0.
sufx := []string{}
hasSingle := false
for _, e := range l {
if len(e.runes) > 1 {
sufx = append(sufx, string(e.runes[1:]))
} else {
hasSingle = true
}
}
if !hasSingle {
b.error(fmt.Errorf("no single entry for starter rune %U found", r))
continue
}
// Unique the suffix set.
sort.Strings(sufx)
key := strings.Join(sufx, "\n")
handle, ok := b.ctHandle[key]
if !ok {
var err error
handle, err = appendTrie(&t.ContractTries, sufx)
if err != nil {
b.error(err)
}
b.ctHandle[key] = handle
}
// Bucket sort entries in index order.
es := make([]*entry, len(l))
for _, e := range l {
var p, sn int
if len(e.runes) > 1 {
str := []byte(string(e.runes[1:]))
p, sn = lookup(&t.ContractTries, handle, str)
if sn != len(str) {
log.Fatalf("%s: processContractions: unexpected length for '%X'; len=%d; want %d", o.id, e.runes, sn, len(str))
}
}
if es[p] != nil {
log.Fatalf("%s: multiple contractions for position %d for rune %U", o.id, p, e.runes[0])
}
es[p] = e
}
// Create collation elements for contractions.
elems := []uint32{}
for _, e := range es {
ce, err := e.encodeBase()
b.errorID(o.id, err)
elems = append(elems, ce)
}
key = fmt.Sprintf("%v", elems)
i, ok := b.ctElem[key]
if !ok {
i = len(t.ContractElem)
b.ctElem[key] = i
t.ContractElem = append(t.ContractElem, elems...)
}
// Store info in entry for starter rune.
es[0].contractionIndex = i
es[0].contractionHandle = handle
}
}

294
vendor/golang.org/x/text/collate/build/colelem.go generated vendored
View File

@ -1,294 +0,0 @@
// 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 build
import (
"fmt"
"unicode"
"golang.org/x/text/internal/colltab"
)
const (
defaultSecondary = 0x20
defaultTertiary = 0x2
maxTertiary = 0x1F
)
type rawCE struct {
w []int
ccc uint8
}
func makeRawCE(w []int, ccc uint8) rawCE {
ce := rawCE{w: make([]int, 4), ccc: ccc}
copy(ce.w, w)
return ce
}
// A collation element is represented as an uint32.
// In the typical case, a rune maps to a single collation element. If a rune
// can be the start of a contraction or expands into multiple collation elements,
// then the collation element that is associated with a rune will have a special
// form to represent such m to n mappings. Such special collation elements
// have a value >= 0x80000000.
const (
maxPrimaryBits = 21
maxSecondaryBits = 12
maxTertiaryBits = 8
)
func makeCE(ce rawCE) (uint32, error) {
v, e := colltab.MakeElem(ce.w[0], ce.w[1], ce.w[2], ce.ccc)
return uint32(v), e
}
// For contractions, collation elements are of the form
// 110bbbbb bbbbbbbb iiiiiiii iiiinnnn, where
// - n* is the size of the first node in the contraction trie.
// - i* is the index of the first node in the contraction trie.
// - b* is the offset into the contraction collation element table.
// See contract.go for details on the contraction trie.
const (
contractID = 0xC0000000
maxNBits = 4
maxTrieIndexBits = 12
maxContractOffsetBits = 13
)
func makeContractIndex(h ctHandle, offset int) (uint32, error) {
if h.n >= 1<<maxNBits {
return 0, fmt.Errorf("size of contraction trie node too large: %d >= %d", h.n, 1<<maxNBits)
}
if h.index >= 1<<maxTrieIndexBits {
return 0, fmt.Errorf("size of contraction trie offset too large: %d >= %d", h.index, 1<<maxTrieIndexBits)
}
if offset >= 1<<maxContractOffsetBits {
return 0, fmt.Errorf("contraction offset out of bounds: %x >= %x", offset, 1<<maxContractOffsetBits)
}
ce := uint32(contractID)
ce += uint32(offset << (maxNBits + maxTrieIndexBits))
ce += uint32(h.index << maxNBits)
ce += uint32(h.n)
return ce, nil
}
// For expansions, collation elements are of the form
// 11100000 00000000 bbbbbbbb bbbbbbbb,
// where b* is the index into the expansion sequence table.
const (
expandID = 0xE0000000
maxExpandIndexBits = 16
)
func makeExpandIndex(index int) (uint32, error) {
if index >= 1<<maxExpandIndexBits {
return 0, fmt.Errorf("expansion index out of bounds: %x >= %x", index, 1<<maxExpandIndexBits)
}
return expandID + uint32(index), nil
}
// Each list of collation elements corresponding to an expansion starts with
// a header indicating the length of the sequence.
func makeExpansionHeader(n int) (uint32, error) {
return uint32(n), nil
}
// Some runes can be expanded using NFKD decomposition. Instead of storing the full
// sequence of collation elements, we decompose the rune and lookup the collation
// elements for each rune in the decomposition and modify the tertiary weights.
// The collation element, in this case, is of the form
// 11110000 00000000 wwwwwwww vvvvvvvv, where
// - v* is the replacement tertiary weight for the first rune,
// - w* is the replacement tertiary weight for the second rune,
// Tertiary weights of subsequent runes should be replaced with maxTertiary.
// See https://www.unicode.org/reports/tr10/#Compatibility_Decompositions for more details.
const (
decompID = 0xF0000000
)
func makeDecompose(t1, t2 int) (uint32, error) {
if t1 >= 256 || t1 < 0 {
return 0, fmt.Errorf("first tertiary weight out of bounds: %d >= 256", t1)
}
if t2 >= 256 || t2 < 0 {
return 0, fmt.Errorf("second tertiary weight out of bounds: %d >= 256", t2)
}
return uint32(t2<<8+t1) + decompID, nil
}
const (
// These constants were taken from https://www.unicode.org/versions/Unicode6.0.0/ch12.pdf.
minUnified rune = 0x4E00
maxUnified = 0x9FFF
minCompatibility = 0xF900
maxCompatibility = 0xFAFF
minRare = 0x3400
maxRare = 0x4DBF
)
const (
commonUnifiedOffset = 0x10000
rareUnifiedOffset = 0x20000 // largest rune in common is U+FAFF
otherOffset = 0x50000 // largest rune in rare is U+2FA1D
illegalOffset = otherOffset + int(unicode.MaxRune)
maxPrimary = illegalOffset + 1
)
// implicitPrimary returns the primary weight for the a rune
// for which there is no entry for the rune in the collation table.
// We take a different approach from the one specified in
// https://unicode.org/reports/tr10/#Implicit_Weights,
// but preserve the resulting relative ordering of the runes.
func implicitPrimary(r rune) int {
if unicode.Is(unicode.Ideographic, r) {
if r >= minUnified && r <= maxUnified {
// The most common case for CJK.
return int(r) + commonUnifiedOffset
}
if r >= minCompatibility && r <= maxCompatibility {
// This will typically not hit. The DUCET explicitly specifies mappings
// for all characters that do not decompose.
return int(r) + commonUnifiedOffset
}
return int(r) + rareUnifiedOffset
}
return int(r) + otherOffset
}
// convertLargeWeights converts collation elements with large
// primaries (either double primaries or for illegal runes)
// to our own representation.
// A CJK character C is represented in the DUCET as
// [.FBxx.0020.0002.C][.BBBB.0000.0000.C]
// We will rewrite these characters to a single CE.
// We assume the CJK values start at 0x8000.
// See https://unicode.org/reports/tr10/#Implicit_Weights
func convertLargeWeights(elems []rawCE) (res []rawCE, err error) {
const (
cjkPrimaryStart = 0xFB40
rarePrimaryStart = 0xFB80
otherPrimaryStart = 0xFBC0
illegalPrimary = 0xFFFE
highBitsMask = 0x3F
lowBitsMask = 0x7FFF
lowBitsFlag = 0x8000
shiftBits = 15
)
for i := 0; i < len(elems); i++ {
ce := elems[i].w
p := ce[0]
if p < cjkPrimaryStart {
continue
}
if p > 0xFFFF {
return elems, fmt.Errorf("found primary weight %X; should be <= 0xFFFF", p)
}
if p >= illegalPrimary {
ce[0] = illegalOffset + p - illegalPrimary
} else {
if i+1 >= len(elems) {
return elems, fmt.Errorf("second part of double primary weight missing: %v", elems)
}
if elems[i+1].w[0]&lowBitsFlag == 0 {
return elems, fmt.Errorf("malformed second part of double primary weight: %v", elems)
}
np := ((p & highBitsMask) << shiftBits) + elems[i+1].w[0]&lowBitsMask
switch {
case p < rarePrimaryStart:
np += commonUnifiedOffset
case p < otherPrimaryStart:
np += rareUnifiedOffset
default:
p += otherOffset
}
ce[0] = np
for j := i + 1; j+1 < len(elems); j++ {
elems[j] = elems[j+1]
}
elems = elems[:len(elems)-1]
}
}
return elems, nil
}
// nextWeight computes the first possible collation weights following elems
// for the given level.
func nextWeight(level colltab.Level, elems []rawCE) []rawCE {
if level == colltab.Identity {
next := make([]rawCE, len(elems))
copy(next, elems)
return next
}
next := []rawCE{makeRawCE(elems[0].w, elems[0].ccc)}
next[0].w[level]++
if level < colltab.Secondary {
next[0].w[colltab.Secondary] = defaultSecondary
}
if level < colltab.Tertiary {
next[0].w[colltab.Tertiary] = defaultTertiary
}
// Filter entries that cannot influence ordering.
for _, ce := range elems[1:] {
skip := true
for i := colltab.Primary; i < level; i++ {
skip = skip && ce.w[i] == 0
}
if !skip {
next = append(next, ce)
}
}
return next
}
func nextVal(elems []rawCE, i int, level colltab.Level) (index, value int) {
for ; i < len(elems) && elems[i].w[level] == 0; i++ {
}
if i < len(elems) {
return i, elems[i].w[level]
}
return i, 0
}
// compareWeights returns -1 if a < b, 1 if a > b, or 0 otherwise.
// It also returns the collation level at which the difference is found.
func compareWeights(a, b []rawCE) (result int, level colltab.Level) {
for level := colltab.Primary; level < colltab.Identity; level++ {
var va, vb int
for ia, ib := 0, 0; ia < len(a) || ib < len(b); ia, ib = ia+1, ib+1 {
ia, va = nextVal(a, ia, level)
ib, vb = nextVal(b, ib, level)
if va != vb {
if va < vb {
return -1, level
} else {
return 1, level
}
}
}
}
return 0, colltab.Identity
}
func equalCE(a, b rawCE) bool {
for i := 0; i < 3; i++ {
if b.w[i] != a.w[i] {
return false
}
}
return true
}
func equalCEArrays(a, b []rawCE) bool {
if len(a) != len(b) {
return false
}
for i := range a {
if !equalCE(a[i], b[i]) {
return false
}
}
return true
}

309
vendor/golang.org/x/text/collate/build/contract.go generated vendored
View File

@ -1,309 +0,0 @@
// 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 build
import (
"fmt"
"io"
"reflect"
"sort"
"strings"
"golang.org/x/text/internal/colltab"
)
// This file contains code for detecting contractions and generating
// the necessary tables.
// Any Unicode Collation Algorithm (UCA) table entry that has more than
// one rune one the left-hand side is called a contraction.
// See https://www.unicode.org/reports/tr10/#Contractions for more details.
//
// We define the following terms:
// initial: a rune that appears as the first rune in a contraction.
// suffix: a sequence of runes succeeding the initial rune
// in a given contraction.
// non-initial: a rune that appears in a suffix.
//
// A rune may be both an initial and a non-initial and may be so in
// many contractions. An initial may typically also appear by itself.
// In case of ambiguities, the UCA requires we match the longest
// contraction.
//
// Many contraction rules share the same set of possible suffixes.
// We store sets of suffixes in a trie that associates an index with
// each suffix in the set. This index can be used to look up a
// collation element associated with the (starter rune, suffix) pair.
//
// The trie is defined on a UTF-8 byte sequence.
// The overall trie is represented as an array of ctEntries. Each node of the trie
// is represented as a subsequence of ctEntries, where each entry corresponds to
// a possible match of a next character in the search string. An entry
// also includes the length and offset to the next sequence of entries
// to check in case of a match.
const (
final = 0
noIndex = 0xFF
)
// ctEntry associates to a matching byte an offset and/or next sequence of
// bytes to check. A ctEntry c is called final if a match means that the
// longest suffix has been found. An entry c is final if c.N == 0.
// A single final entry can match a range of characters to an offset.
// A non-final entry always matches a single byte. Note that a non-final
// entry might still resemble a completed suffix.
// Examples:
// The suffix strings "ab" and "ac" can be represented as:
// []ctEntry{
// {'a', 1, 1, noIndex}, // 'a' by itself does not match, so i is 0xFF.
// {'b', 'c', 0, 1}, // "ab" -> 1, "ac" -> 2
// }
//
// The suffix strings "ab", "abc", "abd", and "abcd" can be represented as:
// []ctEntry{
// {'a', 1, 1, noIndex}, // 'a' must be followed by 'b'.
// {'b', 1, 2, 1}, // "ab" -> 1, may be followed by 'c' or 'd'.
// {'d', 'd', final, 3}, // "abd" -> 3
// {'c', 4, 1, 2}, // "abc" -> 2, may be followed by 'd'.
// {'d', 'd', final, 4}, // "abcd" -> 4
// }
// See genStateTests in contract_test.go for more examples.
type ctEntry struct {
L uint8 // non-final: byte value to match; final: lowest match in range.
H uint8 // non-final: relative index to next block; final: highest match in range.
N uint8 // non-final: length of next block; final: final
I uint8 // result offset. Will be noIndex if more bytes are needed to complete.
}
// contractTrieSet holds a set of contraction tries. The tries are stored
// consecutively in the entry field.
type contractTrieSet []struct{ l, h, n, i uint8 }
// ctHandle is used to identify a trie in the trie set, consisting in an offset
// in the array and the size of the first node.
type ctHandle struct {
index, n int
}
// appendTrie adds a new trie for the given suffixes to the trie set and returns
// a handle to it. The handle will be invalid on error.
func appendTrie(ct *colltab.ContractTrieSet, suffixes []string) (ctHandle, error) {
es := make([]stridx, len(suffixes))
for i, s := range suffixes {
es[i].str = s
}
sort.Sort(offsetSort(es))
for i := range es {
es[i].index = i + 1
}
sort.Sort(genidxSort(es))
i := len(*ct)
n, err := genStates(ct, es)
if err != nil {
*ct = (*ct)[:i]
return ctHandle{}, err
}
return ctHandle{i, n}, nil
}
// genStates generates ctEntries for a given suffix set and returns
// the number of entries for the first node.
func genStates(ct *colltab.ContractTrieSet, sis []stridx) (int, error) {
if len(sis) == 0 {
return 0, fmt.Errorf("genStates: list of suffices must be non-empty")
}
start := len(*ct)
// create entries for differing first bytes.
for _, si := range sis {
s := si.str
if len(s) == 0 {
continue
}
added := false
c := s[0]
if len(s) > 1 {
for j := len(*ct) - 1; j >= start; j-- {
if (*ct)[j].L == c {
added = true
break
}
}
if !added {
*ct = append(*ct, ctEntry{L: c, I: noIndex})
}
} else {
for j := len(*ct) - 1; j >= start; j-- {
// Update the offset for longer suffixes with the same byte.
if (*ct)[j].L == c {
(*ct)[j].I = uint8(si.index)
added = true
}
// Extend range of final ctEntry, if possible.
if (*ct)[j].H+1 == c {
(*ct)[j].H = c
added = true
}
}
if !added {
*ct = append(*ct, ctEntry{L: c, H: c, N: final, I: uint8(si.index)})
}
}
}
n := len(*ct) - start
// Append nodes for the remainder of the suffixes for each ctEntry.
sp := 0
for i, end := start, len(*ct); i < end; i++ {
fe := (*ct)[i]
if fe.H == 0 { // uninitialized non-final
ln := len(*ct) - start - n
if ln > 0xFF {
return 0, fmt.Errorf("genStates: relative block offset too large: %d > 255", ln)
}
fe.H = uint8(ln)
// Find first non-final strings with same byte as current entry.
for ; sis[sp].str[0] != fe.L; sp++ {
}
se := sp + 1
for ; se < len(sis) && len(sis[se].str) > 1 && sis[se].str[0] == fe.L; se++ {
}
sl := sis[sp:se]
sp = se
for i, si := range sl {
sl[i].str = si.str[1:]
}
nn, err := genStates(ct, sl)
if err != nil {
return 0, err
}
fe.N = uint8(nn)
(*ct)[i] = fe
}
}
sort.Sort(entrySort((*ct)[start : start+n]))
return n, nil
}
// There may be both a final and non-final entry for a byte if the byte
// is implied in a range of matches in the final entry.
// We need to ensure that the non-final entry comes first in that case.
type entrySort colltab.ContractTrieSet
func (fe entrySort) Len() int { return len(fe) }
func (fe entrySort) Swap(i, j int) { fe[i], fe[j] = fe[j], fe[i] }
func (fe entrySort) Less(i, j int) bool {
return fe[i].L > fe[j].L
}
// stridx is used for sorting suffixes and their associated offsets.
type stridx struct {
str string
index int
}
// For computing the offsets, we first sort by size, and then by string.
// This ensures that strings that only differ in the last byte by 1
// are sorted consecutively in increasing order such that they can
// be packed as a range in a final ctEntry.
type offsetSort []stridx
func (si offsetSort) Len() int { return len(si) }
func (si offsetSort) Swap(i, j int) { si[i], si[j] = si[j], si[i] }
func (si offsetSort) Less(i, j int) bool {
if len(si[i].str) != len(si[j].str) {
return len(si[i].str) > len(si[j].str)
}
return si[i].str < si[j].str
}
// For indexing, we want to ensure that strings are sorted in string order, where
// for strings with the same prefix, we put longer strings before shorter ones.
type genidxSort []stridx
func (si genidxSort) Len() int { return len(si) }
func (si genidxSort) Swap(i, j int) { si[i], si[j] = si[j], si[i] }
func (si genidxSort) Less(i, j int) bool {
if strings.HasPrefix(si[j].str, si[i].str) {
return false
}
if strings.HasPrefix(si[i].str, si[j].str) {
return true
}
return si[i].str < si[j].str
}
// lookup matches the longest suffix in str and returns the associated offset
// and the number of bytes consumed.
func lookup(ct *colltab.ContractTrieSet, h ctHandle, str []byte) (index, ns int) {
states := (*ct)[h.index:]
p := 0
n := h.n
for i := 0; i < n && p < len(str); {
e := states[i]
c := str[p]
if c >= e.L {
if e.L == c {
p++
if e.I != noIndex {
index, ns = int(e.I), p
}
if e.N != final {
// set to new state
i, states, n = 0, states[int(e.H)+n:], int(e.N)
} else {
return
}
continue
} else if e.N == final && c <= e.H {
p++
return int(c-e.L) + int(e.I), p
}
}
i++
}
return
}
// print writes the contractTrieSet t as compilable Go code to w. It returns
// the total number of bytes written and the size of the resulting data structure in bytes.
func print(t *colltab.ContractTrieSet, w io.Writer, name string) (n, size int, err error) {
update3 := func(nn, sz int, e error) {
n += nn
if err == nil {
err = e
}
size += sz
}
update2 := func(nn int, e error) { update3(nn, 0, e) }
update3(printArray(*t, w, name))
update2(fmt.Fprintf(w, "var %sContractTrieSet = ", name))
update3(printStruct(*t, w, name))
update2(fmt.Fprintln(w))
return
}
func printArray(ct colltab.ContractTrieSet, w io.Writer, name string) (n, size int, err error) {
p := func(f string, a ...interface{}) {
nn, e := fmt.Fprintf(w, f, a...)
n += nn
if err == nil {
err = e
}
}
size = len(ct) * 4
p("// %sCTEntries: %d entries, %d bytes\n", name, len(ct), size)
p("var %sCTEntries = [%d]struct{L,H,N,I uint8}{\n", name, len(ct))
for _, fe := range ct {
p("\t{0x%X, 0x%X, %d, %d},\n", fe.L, fe.H, fe.N, fe.I)
}
p("}\n")
return
}
func printStruct(ct colltab.ContractTrieSet, w io.Writer, name string) (n, size int, err error) {
n, err = fmt.Fprintf(w, "colltab.ContractTrieSet( %sCTEntries[:] )", name)
size = int(reflect.TypeOf(ct).Size())
return
}

393
vendor/golang.org/x/text/collate/build/order.go generated vendored
View File

@ -1,393 +0,0 @@
// 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 build
import (
"fmt"
"log"
"sort"
"strings"
"unicode"
"golang.org/x/text/internal/colltab"
"golang.org/x/text/unicode/norm"
)
type logicalAnchor int
const (
firstAnchor logicalAnchor = -1
noAnchor = 0
lastAnchor = 1
)
// entry is used to keep track of a single entry in the collation element table
// during building. Examples of entries can be found in the Default Unicode
// Collation Element Table.
// See https://www.unicode.org/Public/UCA/6.0.0/allkeys.txt.
type entry struct {
str string // same as string(runes)
runes []rune
elems []rawCE // the collation elements
extend string // weights of extend to be appended to elems
before bool // weights relative to next instead of previous.
lock bool // entry is used in extension and can no longer be moved.
// prev, next, and level are used to keep track of tailorings.
prev, next *entry
level colltab.Level // next differs at this level
skipRemove bool // do not unlink when removed
decompose bool // can use NFKD decomposition to generate elems
exclude bool // do not include in table
implicit bool // derived, is not included in the list
modified bool // entry was modified in tailoring
logical logicalAnchor
expansionIndex int // used to store index into expansion table
contractionHandle ctHandle
contractionIndex int // index into contraction elements
}
func (e *entry) String() string {
return fmt.Sprintf("%X (%q) -> %X (ch:%x; ci:%d, ei:%d)",
e.runes, e.str, e.elems, e.contractionHandle, e.contractionIndex, e.expansionIndex)
}
func (e *entry) skip() bool {
return e.contraction()
}
func (e *entry) expansion() bool {
return !e.decompose && len(e.elems) > 1
}
func (e *entry) contraction() bool {
return len(e.runes) > 1
}
func (e *entry) contractionStarter() bool {
return e.contractionHandle.n != 0
}
// nextIndexed gets the next entry that needs to be stored in the table.
// It returns the entry and the collation level at which the next entry differs
// from the current entry.
// Entries that can be explicitly derived and logical reset positions are
// examples of entries that will not be indexed.
func (e *entry) nextIndexed() (*entry, colltab.Level) {
level := e.level
for e = e.next; e != nil && (e.exclude || len(e.elems) == 0); e = e.next {
if e.level < level {
level = e.level
}
}
return e, level
}
// remove unlinks entry e from the sorted chain and clears the collation
// elements. e may not be at the front or end of the list. This should always
// be the case, as the front and end of the list are always logical anchors,
// which may not be removed.
func (e *entry) remove() {
if e.logical != noAnchor {
log.Fatalf("may not remove anchor %q", e.str)
}
// TODO: need to set e.prev.level to e.level if e.level is smaller?
e.elems = nil
if !e.skipRemove {
if e.prev != nil {
e.prev.next = e.next
}
if e.next != nil {
e.next.prev = e.prev
}
}
e.skipRemove = false
}
// insertAfter inserts n after e.
func (e *entry) insertAfter(n *entry) {
if e == n {
panic("e == anchor")
}
if e == nil {
panic("unexpected nil anchor")
}
n.remove()
n.decompose = false // redo decomposition test
n.next = e.next
n.prev = e
if e.next != nil {
e.next.prev = n
}
e.next = n
}
// insertBefore inserts n before e.
func (e *entry) insertBefore(n *entry) {
if e == n {
panic("e == anchor")
}
if e == nil {
panic("unexpected nil anchor")
}
n.remove()
n.decompose = false // redo decomposition test
n.prev = e.prev
n.next = e
if e.prev != nil {
e.prev.next = n
}
e.prev = n
}
func (e *entry) encodeBase() (ce uint32, err error) {
switch {
case e.expansion():
ce, err = makeExpandIndex(e.expansionIndex)
default:
if e.decompose {
log.Fatal("decompose should be handled elsewhere")
}
ce, err = makeCE(e.elems[0])
}
return
}
func (e *entry) encode() (ce uint32, err error) {
if e.skip() {
log.Fatal("cannot build colElem for entry that should be skipped")
}
switch {
case e.decompose:
t1 := e.elems[0].w[2]
t2 := 0
if len(e.elems) > 1 {
t2 = e.elems[1].w[2]
}
ce, err = makeDecompose(t1, t2)
case e.contractionStarter():
ce, err = makeContractIndex(e.contractionHandle, e.contractionIndex)
default:
if len(e.runes) > 1 {
log.Fatal("colElem: contractions are handled in contraction trie")
}
ce, err = e.encodeBase()
}
return
}
// entryLess returns true if a sorts before b and false otherwise.
func entryLess(a, b *entry) bool {
if res, _ := compareWeights(a.elems, b.elems); res != 0 {
return res == -1
}
if a.logical != noAnchor {
return a.logical == firstAnchor
}
if b.logical != noAnchor {
return b.logical == lastAnchor
}
return a.str < b.str
}
type sortedEntries []*entry
func (s sortedEntries) Len() int {
return len(s)
}
func (s sortedEntries) Swap(i, j int) {
s[i], s[j] = s[j], s[i]
}
func (s sortedEntries) Less(i, j int) bool {
return entryLess(s[i], s[j])
}
type ordering struct {
id string
entryMap map[string]*entry
ordered []*entry
handle *trieHandle
}
// insert inserts e into both entryMap and ordered.
// Note that insert simply appends e to ordered. To reattain a sorted
// order, o.sort() should be called.
func (o *ordering) insert(e *entry) {
if e.logical == noAnchor {
o.entryMap[e.str] = e
} else {
// Use key format as used in UCA rules.
o.entryMap[fmt.Sprintf("[%s]", e.str)] = e
// Also add index entry for XML format.
o.entryMap[fmt.Sprintf("<%s/>", strings.Replace(e.str, " ", "_", -1))] = e
}
o.ordered = append(o.ordered, e)
}
// newEntry creates a new entry for the given info and inserts it into
// the index.
func (o *ordering) newEntry(s string, ces []rawCE) *entry {
e := &entry{
runes: []rune(s),
elems: ces,
str: s,
}
o.insert(e)
return e
}
// find looks up and returns the entry for the given string.
// It returns nil if str is not in the index and if an implicit value
// cannot be derived, that is, if str represents more than one rune.
func (o *ordering) find(str string) *entry {
e := o.entryMap[str]
if e == nil {
r := []rune(str)
if len(r) == 1 {
const (
firstHangul = 0xAC00
lastHangul = 0xD7A3
)
if r[0] >= firstHangul && r[0] <= lastHangul {
ce := []rawCE{}
nfd := norm.NFD.String(str)
for _, r := range nfd {
ce = append(ce, o.find(string(r)).elems...)
}
e = o.newEntry(nfd, ce)
} else {
e = o.newEntry(string(r[0]), []rawCE{
{w: []int{
implicitPrimary(r[0]),
defaultSecondary,
defaultTertiary,
int(r[0]),
},
},
})
e.modified = true
}
e.exclude = true // do not index implicits
}
}
return e
}
// makeRootOrdering returns a newly initialized ordering value and populates
// it with a set of logical reset points that can be used as anchors.
// The anchors first_tertiary_ignorable and __END__ will always sort at
// the beginning and end, respectively. This means that prev and next are non-nil
// for any indexed entry.
func makeRootOrdering() ordering {
const max = unicode.MaxRune
o := ordering{
entryMap: make(map[string]*entry),
}
insert := func(typ logicalAnchor, s string, ce []int) {
e := &entry{
elems: []rawCE{{w: ce}},
str: s,
exclude: true,
logical: typ,
}
o.insert(e)
}
insert(firstAnchor, "first tertiary ignorable", []int{0, 0, 0, 0})
insert(lastAnchor, "last tertiary ignorable", []int{0, 0, 0, max})
insert(lastAnchor, "last primary ignorable", []int{0, defaultSecondary, defaultTertiary, max})
insert(lastAnchor, "last non ignorable", []int{maxPrimary, defaultSecondary, defaultTertiary, max})
insert(lastAnchor, "__END__", []int{1 << maxPrimaryBits, defaultSecondary, defaultTertiary, max})
return o
}
// patchForInsert eleminates entries from the list with more than one collation element.
// The next and prev fields of the eliminated entries still point to appropriate
// values in the newly created list.
// It requires that sort has been called.
func (o *ordering) patchForInsert() {
for i := 0; i < len(o.ordered)-1; {
e := o.ordered[i]
lev := e.level
n := e.next
for ; n != nil && len(n.elems) > 1; n = n.next {
if n.level < lev {
lev = n.level
}
n.skipRemove = true
}
for ; o.ordered[i] != n; i++ {
o.ordered[i].level = lev
o.ordered[i].next = n
o.ordered[i+1].prev = e
}
}
}
// clone copies all ordering of es into a new ordering value.
func (o *ordering) clone() *ordering {
o.sort()
oo := ordering{
entryMap: make(map[string]*entry),
}
for _, e := range o.ordered {
ne := &entry{
runes: e.runes,
elems: e.elems,
str: e.str,
decompose: e.decompose,
exclude: e.exclude,
logical: e.logical,
}
oo.insert(ne)
}
oo.sort() // link all ordering.
oo.patchForInsert()
return &oo
}
// front returns the first entry to be indexed.
// It assumes that sort() has been called.
func (o *ordering) front() *entry {
e := o.ordered[0]
if e.prev != nil {
log.Panicf("unexpected first entry: %v", e)
}
// The first entry is always a logical position, which should not be indexed.
e, _ = e.nextIndexed()
return e
}
// sort sorts all ordering based on their collation elements and initializes
// the prev, next, and level fields accordingly.
func (o *ordering) sort() {
sort.Sort(sortedEntries(o.ordered))
l := o.ordered
for i := 1; i < len(l); i++ {
k := i - 1
l[k].next = l[i]
_, l[k].level = compareWeights(l[k].elems, l[i].elems)
l[i].prev = l[k]
}
}
// genColElems generates a collation element array from the runes in str. This
// assumes that all collation elements have already been added to the Builder.
func (o *ordering) genColElems(str string) []rawCE {
elems := []rawCE{}
for _, r := range []rune(str) {
for _, ce := range o.find(string(r)).elems {
if ce.w[0] != 0 || ce.w[1] != 0 || ce.w[2] != 0 {
elems = append(elems, ce)
}
}
}
return elems
}

81
vendor/golang.org/x/text/collate/build/table.go generated vendored
View File

@ -1,81 +0,0 @@
// 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 build
import (
"fmt"
"io"
"reflect"
"golang.org/x/text/internal/colltab"
)
// table is an intermediate structure that roughly resembles the table in collate.
type table struct {
colltab.Table
trie trie
root *trieHandle
}
// print writes the table as Go compilable code to w. It prefixes the
// variable names with name. It returns the number of bytes written
// and the size of the resulting table.
func (t *table) fprint(w io.Writer, name string) (n, size int, err error) {
update := func(nn, sz int, e error) {
n += nn
if err == nil {
err = e
}
size += sz
}
// Write arrays needed for the structure.
update(printColElems(w, t.ExpandElem, name+"ExpandElem"))
update(printColElems(w, t.ContractElem, name+"ContractElem"))
update(t.trie.printArrays(w, name))
update(printArray(t.ContractTries, w, name))
nn, e := fmt.Fprintf(w, "// Total size of %sTable is %d bytes\n", name, size)
update(nn, 0, e)
return
}
func (t *table) fprintIndex(w io.Writer, h *trieHandle, id string) (n int, err error) {
p := func(f string, a ...interface{}) {
nn, e := fmt.Fprintf(w, f, a...)
n += nn
if err == nil {
err = e
}
}
p("\t{ // %s\n", id)
p("\t\tlookupOffset: 0x%x,\n", h.lookupStart)
p("\t\tvaluesOffset: 0x%x,\n", h.valueStart)
p("\t},\n")
return
}
func printColElems(w io.Writer, a []uint32, name string) (n, sz int, err error) {
p := func(f string, a ...interface{}) {
nn, e := fmt.Fprintf(w, f, a...)
n += nn
if err == nil {
err = e
}
}
sz = len(a) * int(reflect.TypeOf(uint32(0)).Size())
p("// %s: %d entries, %d bytes\n", name, len(a), sz)
p("var %s = [%d]uint32 {", name, len(a))
for i, c := range a {
switch {
case i%64 == 0:
p("\n\t// Block %d, offset 0x%x\n", i/64, i)
case (i%64)%6 == 0:
p("\n\t")
}
p("0x%.8X, ", c)
}
p("\n}\n\n")
return
}

290
vendor/golang.org/x/text/collate/build/trie.go generated vendored
View File

@ -1,290 +0,0 @@
// 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.
// The trie in this file is used to associate the first full character
// in a UTF-8 string to a collation element.
// All but the last byte in a UTF-8 byte sequence are
// used to look up offsets in the index table to be used for the next byte.
// The last byte is used to index into a table of collation elements.
// This file contains the code for the generation of the trie.
package build
import (
"fmt"
"hash/fnv"
"io"
"reflect"
)
const (
blockSize = 64
blockOffset = 2 // Subtract 2 blocks to compensate for the 0x80 added to continuation bytes.
)
type trieHandle struct {
lookupStart uint16 // offset in table for first byte
valueStart uint16 // offset in table for first byte
}
type trie struct {
index []uint16
values []uint32
}
// trieNode is the intermediate trie structure used for generating a trie.
type trieNode struct {
index []*trieNode
value []uint32
b byte
refValue uint16
refIndex uint16
}
func newNode() *trieNode {
return &trieNode{
index: make([]*trieNode, 64),
value: make([]uint32, 128), // root node size is 128 instead of 64
}
}
func (n *trieNode) isInternal() bool {
return n.value != nil
}
func (n *trieNode) insert(r rune, value uint32) {
const maskx = 0x3F // mask out two most-significant bits
str := string(r)
if len(str) == 1 {
n.value[str[0]] = value
return
}
for i := 0; i < len(str)-1; i++ {
b := str[i] & maskx
if n.index == nil {
n.index = make([]*trieNode, blockSize)
}
nn := n.index[b]
if nn == nil {
nn = &trieNode{}
nn.b = b
n.index[b] = nn
}
n = nn
}
if n.value == nil {
n.value = make([]uint32, blockSize)
}
b := str[len(str)-1] & maskx
n.value[b] = value
}
type trieBuilder struct {
t *trie
roots []*trieHandle
lookupBlocks []*trieNode
valueBlocks []*trieNode
lookupBlockIdx map[uint32]*trieNode
valueBlockIdx map[uint32]*trieNode
}
func newTrieBuilder() *trieBuilder {
index := &trieBuilder{}
index.lookupBlocks = make([]*trieNode, 0)
index.valueBlocks = make([]*trieNode, 0)
index.lookupBlockIdx = make(map[uint32]*trieNode)
index.valueBlockIdx = make(map[uint32]*trieNode)
// The third nil is the default null block. The other two blocks
// are used to guarantee an offset of at least 3 for each block.
index.lookupBlocks = append(index.lookupBlocks, nil, nil, nil)
index.t = &trie{}
return index
}
func (b *trieBuilder) computeOffsets(n *trieNode) *trieNode {
hasher := fnv.New32()
if n.index != nil {
for i, nn := range n.index {
var vi, vv uint16
if nn != nil {
nn = b.computeOffsets(nn)
n.index[i] = nn
vi = nn.refIndex
vv = nn.refValue
}
hasher.Write([]byte{byte(vi >> 8), byte(vi)})
hasher.Write([]byte{byte(vv >> 8), byte(vv)})
}
h := hasher.Sum32()
nn, ok := b.lookupBlockIdx[h]
if !ok {
n.refIndex = uint16(len(b.lookupBlocks)) - blockOffset
b.lookupBlocks = append(b.lookupBlocks, n)
b.lookupBlockIdx[h] = n
} else {
n = nn
}
} else {
for _, v := range n.value {
hasher.Write([]byte{byte(v >> 24), byte(v >> 16), byte(v >> 8), byte(v)})
}
h := hasher.Sum32()
nn, ok := b.valueBlockIdx[h]
if !ok {
n.refValue = uint16(len(b.valueBlocks)) - blockOffset
n.refIndex = n.refValue
b.valueBlocks = append(b.valueBlocks, n)
b.valueBlockIdx[h] = n
} else {
n = nn
}
}
return n
}
func (b *trieBuilder) addStartValueBlock(n *trieNode) uint16 {
hasher := fnv.New32()
for _, v := range n.value[:2*blockSize] {
hasher.Write([]byte{byte(v >> 24), byte(v >> 16), byte(v >> 8), byte(v)})
}
h := hasher.Sum32()
nn, ok := b.valueBlockIdx[h]
if !ok {
n.refValue = uint16(len(b.valueBlocks))
n.refIndex = n.refValue
b.valueBlocks = append(b.valueBlocks, n)
// Add a dummy block to accommodate the double block size.
b.valueBlocks = append(b.valueBlocks, nil)
b.valueBlockIdx[h] = n
} else {
n = nn
}
return n.refValue
}
func genValueBlock(t *trie, n *trieNode) {
if n != nil {
for _, v := range n.value {
t.values = append(t.values, v)
}
}
}
func genLookupBlock(t *trie, n *trieNode) {
for _, nn := range n.index {
v := uint16(0)
if nn != nil {
if n.index != nil {
v = nn.refIndex
} else {
v = nn.refValue
}
}
t.index = append(t.index, v)
}
}
func (b *trieBuilder) addTrie(n *trieNode) *trieHandle {
h := &trieHandle{}
b.roots = append(b.roots, h)
h.valueStart = b.addStartValueBlock(n)
if len(b.roots) == 1 {
// We insert a null block after the first start value block.
// This ensures that continuation bytes UTF-8 sequences of length
// greater than 2 will automatically hit a null block if there
// was an undefined entry.
b.valueBlocks = append(b.valueBlocks, nil)
}
n = b.computeOffsets(n)
// Offset by one extra block as the first byte starts at 0xC0 instead of 0x80.
h.lookupStart = n.refIndex - 1
return h
}
// generate generates and returns the trie for n.
func (b *trieBuilder) generate() (t *trie, err error) {
t = b.t
if len(b.valueBlocks) >= 1<<16 {
return nil, fmt.Errorf("maximum number of value blocks exceeded (%d > %d)", len(b.valueBlocks), 1<<16)
}
if len(b.lookupBlocks) >= 1<<16 {
return nil, fmt.Errorf("maximum number of lookup blocks exceeded (%d > %d)", len(b.lookupBlocks), 1<<16)
}
genValueBlock(t, b.valueBlocks[0])
genValueBlock(t, &trieNode{value: make([]uint32, 64)})
for i := 2; i < len(b.valueBlocks); i++ {
genValueBlock(t, b.valueBlocks[i])
}
n := &trieNode{index: make([]*trieNode, 64)}
genLookupBlock(t, n)
genLookupBlock(t, n)
genLookupBlock(t, n)
for i := 3; i < len(b.lookupBlocks); i++ {
genLookupBlock(t, b.lookupBlocks[i])
}
return b.t, nil
}
func (t *trie) printArrays(w io.Writer, name string) (n, size int, err error) {
p := func(f string, a ...interface{}) {
nn, e := fmt.Fprintf(w, f, a...)
n += nn
if err == nil {
err = e
}
}
nv := len(t.values)
p("// %sValues: %d entries, %d bytes\n", name, nv, nv*4)
p("// Block 2 is the null block.\n")
p("var %sValues = [%d]uint32 {", name, nv)
var printnewline bool
for i, v := range t.values {
if i%blockSize == 0 {
p("\n\t// Block %#x, offset %#x", i/blockSize, i)
}
if i%4 == 0 {
printnewline = true
}
if v != 0 {
if printnewline {
p("\n\t")
printnewline = false
}
p("%#04x:%#08x, ", i, v)
}
}
p("\n}\n\n")
ni := len(t.index)
p("// %sLookup: %d entries, %d bytes\n", name, ni, ni*2)
p("// Block 0 is the null block.\n")
p("var %sLookup = [%d]uint16 {", name, ni)
printnewline = false
for i, v := range t.index {
if i%blockSize == 0 {
p("\n\t// Block %#x, offset %#x", i/blockSize, i)
}
if i%8 == 0 {
printnewline = true
}
if v != 0 {
if printnewline {
p("\n\t")
printnewline = false
}
p("%#03x:%#02x, ", i, v)
}
}
p("\n}\n\n")
return n, nv*4 + ni*2, err
}
func (t *trie) printStruct(w io.Writer, handle *trieHandle, name string) (n, sz int, err error) {
const msg = "trie{ %sLookup[%d:], %sValues[%d:], %sLookup[:], %sValues[:]}"
n, err = fmt.Fprintf(w, msg, name, handle.lookupStart*blockSize, name, handle.valueStart*blockSize, name, name)
sz += int(reflect.TypeOf(trie{}).Size())
return
}

403
vendor/golang.org/x/text/collate/collate.go generated vendored
View File

@ -1,403 +0,0 @@
// 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.
// TODO: remove hard-coded versions when we have implemented fractional weights.
// The current implementation is incompatible with later CLDR versions.
//go:generate go run maketables.go -cldr=23 -unicode=6.2.0
// Package collate contains types for comparing and sorting Unicode strings
// according to a given collation order.
package collate // import "golang.org/x/text/collate"
import (
"bytes"
"strings"
"golang.org/x/text/internal/colltab"
"golang.org/x/text/language"
)
// Collator provides functionality for comparing strings for a given
// collation order.
type Collator struct {
options
sorter sorter
_iter [2]iter
}
func (c *Collator) iter(i int) *iter {
// TODO: evaluate performance for making the second iterator optional.
return &c._iter[i]
}
// Supported returns the list of languages for which collating differs from its parent.
func Supported() []language.Tag {
// TODO: use language.Coverage instead.
t := make([]language.Tag, len(tags))
copy(t, tags)
return t
}
func init() {
ids := strings.Split(availableLocales, ",")
tags = make([]language.Tag, len(ids))
for i, s := range ids {
tags[i] = language.Raw.MustParse(s)
}
}
var tags []language.Tag
// New returns a new Collator initialized for the given locale.
func New(t language.Tag, o ...Option) *Collator {
index := colltab.MatchLang(t, tags)
c := newCollator(getTable(locales[index]))
// Set options from the user-supplied tag.
c.setFromTag(t)
// Set the user-supplied options.
c.setOptions(o)
c.init()
return c
}
// NewFromTable returns a new Collator for the given Weighter.
func NewFromTable(w colltab.Weighter, o ...Option) *Collator {
c := newCollator(w)
c.setOptions(o)
c.init()
return c
}
func (c *Collator) init() {
if c.numeric {
c.t = colltab.NewNumericWeighter(c.t)
}
c._iter[0].init(c)
c._iter[1].init(c)
}
// Buffer holds keys generated by Key and KeyString.
type Buffer struct {
buf [4096]byte
key []byte
}
func (b *Buffer) init() {
if b.key == nil {
b.key = b.buf[:0]
}
}
// Reset clears the buffer from previous results generated by Key and KeyString.
func (b *Buffer) Reset() {
b.key = b.key[:0]
}
// Compare returns an integer comparing the two byte slices.
// The result will be 0 if a==b, -1 if a < b, and +1 if a > b.
func (c *Collator) Compare(a, b []byte) int {
// TODO: skip identical prefixes once we have a fast way to detect if a rune is
// part of a contraction. This would lead to roughly a 10% speedup for the colcmp regtest.
c.iter(0).SetInput(a)
c.iter(1).SetInput(b)
if res := c.compare(); res != 0 {
return res
}
if !c.ignore[colltab.Identity] {
return bytes.Compare(a, b)
}
return 0
}
// CompareString returns an integer comparing the two strings.
// The result will be 0 if a==b, -1 if a < b, and +1 if a > b.
func (c *Collator) CompareString(a, b string) int {
// TODO: skip identical prefixes once we have a fast way to detect if a rune is
// part of a contraction. This would lead to roughly a 10% speedup for the colcmp regtest.
c.iter(0).SetInputString(a)
c.iter(1).SetInputString(b)
if res := c.compare(); res != 0 {
return res
}
if !c.ignore[colltab.Identity] {
if a < b {
return -1
} else if a > b {
return 1
}
}
return 0
}
func compareLevel(f func(i *iter) int, a, b *iter) int {
a.pce = 0
b.pce = 0
for {
va := f(a)
vb := f(b)
if va != vb {
if va < vb {
return -1
}
return 1
} else if va == 0 {
break
}
}
return 0
}
func (c *Collator) compare() int {
ia, ib := c.iter(0), c.iter(1)
// Process primary level
if c.alternate != altShifted {
// TODO: implement script reordering
if res := compareLevel((*iter).nextPrimary, ia, ib); res != 0 {
return res
}
} else {
// TODO: handle shifted
}
if !c.ignore[colltab.Secondary] {
f := (*iter).nextSecondary
if c.backwards {
f = (*iter).prevSecondary
}
if res := compareLevel(f, ia, ib); res != 0 {
return res
}
}
// TODO: special case handling (Danish?)
if !c.ignore[colltab.Tertiary] || c.caseLevel {
if res := compareLevel((*iter).nextTertiary, ia, ib); res != 0 {
return res
}
if !c.ignore[colltab.Quaternary] {
if res := compareLevel((*iter).nextQuaternary, ia, ib); res != 0 {
return res
}
}
}
return 0
}
// Key returns the collation key for str.
// Passing the buffer buf may avoid memory allocations.
// The returned slice will point to an allocation in Buffer and will remain
// valid until the next call to buf.Reset().
func (c *Collator) Key(buf *Buffer, str []byte) []byte {
// See https://www.unicode.org/reports/tr10/#Main_Algorithm for more details.
buf.init()
return c.key(buf, c.getColElems(str))
}
// KeyFromString returns the collation key for str.
// Passing the buffer buf may avoid memory allocations.
// The returned slice will point to an allocation in Buffer and will retain
// valid until the next call to buf.ResetKeys().
func (c *Collator) KeyFromString(buf *Buffer, str string) []byte {
// See https://www.unicode.org/reports/tr10/#Main_Algorithm for more details.
buf.init()
return c.key(buf, c.getColElemsString(str))
}
func (c *Collator) key(buf *Buffer, w []colltab.Elem) []byte {
processWeights(c.alternate, c.t.Top(), w)
kn := len(buf.key)
c.keyFromElems(buf, w)
return buf.key[kn:]
}
func (c *Collator) getColElems(str []byte) []colltab.Elem {
i := c.iter(0)
i.SetInput(str)
for i.Next() {
}
return i.Elems
}
func (c *Collator) getColElemsString(str string) []colltab.Elem {
i := c.iter(0)
i.SetInputString(str)
for i.Next() {
}
return i.Elems
}
type iter struct {
wa [512]colltab.Elem
colltab.Iter
pce int
}
func (i *iter) init(c *Collator) {
i.Weighter = c.t
i.Elems = i.wa[:0]
}
func (i *iter) nextPrimary() int {
for {
for ; i.pce < i.N; i.pce++ {
if v := i.Elems[i.pce].Primary(); v != 0 {
i.pce++
return v
}
}
if !i.Next() {
return 0
}
}
panic("should not reach here")
}
func (i *iter) nextSecondary() int {
for ; i.pce < len(i.Elems); i.pce++ {
if v := i.Elems[i.pce].Secondary(); v != 0 {
i.pce++
return v
}
}
return 0
}
func (i *iter) prevSecondary() int {
for ; i.pce < len(i.Elems); i.pce++ {
if v := i.Elems[len(i.Elems)-i.pce-1].Secondary(); v != 0 {
i.pce++
return v
}
}
return 0
}
func (i *iter) nextTertiary() int {
for ; i.pce < len(i.Elems); i.pce++ {
if v := i.Elems[i.pce].Tertiary(); v != 0 {
i.pce++
return int(v)
}
}
return 0
}
func (i *iter) nextQuaternary() int {
for ; i.pce < len(i.Elems); i.pce++ {
if v := i.Elems[i.pce].Quaternary(); v != 0 {
i.pce++
return v
}
}
return 0
}
func appendPrimary(key []byte, p int) []byte {
// Convert to variable length encoding; supports up to 23 bits.
if p <= 0x7FFF {
key = append(key, uint8(p>>8), uint8(p))
} else {
key = append(key, uint8(p>>16)|0x80, uint8(p>>8), uint8(p))
}
return key
}
// keyFromElems converts the weights ws to a compact sequence of bytes.
// The result will be appended to the byte buffer in buf.
func (c *Collator) keyFromElems(buf *Buffer, ws []colltab.Elem) {
for _, v := range ws {
if w := v.Primary(); w > 0 {
buf.key = appendPrimary(buf.key, w)
}
}
if !c.ignore[colltab.Secondary] {
buf.key = append(buf.key, 0, 0)
// TODO: we can use one 0 if we can guarantee that all non-zero weights are > 0xFF.
if !c.backwards {
for _, v := range ws {
if w := v.Secondary(); w > 0 {
buf.key = append(buf.key, uint8(w>>8), uint8(w))
}
}
} else {
for i := len(ws) - 1; i >= 0; i-- {
if w := ws[i].Secondary(); w > 0 {
buf.key = append(buf.key, uint8(w>>8), uint8(w))
}
}
}
} else if c.caseLevel {
buf.key = append(buf.key, 0, 0)
}
if !c.ignore[colltab.Tertiary] || c.caseLevel {
buf.key = append(buf.key, 0, 0)
for _, v := range ws {
if w := v.Tertiary(); w > 0 {
buf.key = append(buf.key, uint8(w))
}
}
// Derive the quaternary weights from the options and other levels.
// Note that we represent MaxQuaternary as 0xFF. The first byte of the
// representation of a primary weight is always smaller than 0xFF,
// so using this single byte value will compare correctly.
if !c.ignore[colltab.Quaternary] && c.alternate >= altShifted {
if c.alternate == altShiftTrimmed {
lastNonFFFF := len(buf.key)
buf.key = append(buf.key, 0)
for _, v := range ws {
if w := v.Quaternary(); w == colltab.MaxQuaternary {
buf.key = append(buf.key, 0xFF)
} else if w > 0 {
buf.key = appendPrimary(buf.key, w)
lastNonFFFF = len(buf.key)
}
}
buf.key = buf.key[:lastNonFFFF]
} else {
buf.key = append(buf.key, 0)
for _, v := range ws {
if w := v.Quaternary(); w == colltab.MaxQuaternary {
buf.key = append(buf.key, 0xFF)
} else if w > 0 {
buf.key = appendPrimary(buf.key, w)
}
}
}
}
}
}
func processWeights(vw alternateHandling, top uint32, wa []colltab.Elem) {
ignore := false
vtop := int(top)
switch vw {
case altShifted, altShiftTrimmed:
for i := range wa {
if p := wa[i].Primary(); p <= vtop && p != 0 {
wa[i] = colltab.MakeQuaternary(p)
ignore = true
} else if p == 0 {
if ignore {
wa[i] = colltab.Ignore
}
} else {
ignore = false
}
}
case altBlanked:
for i := range wa {
if p := wa[i].Primary(); p <= vtop && (ignore || p != 0) {
wa[i] = colltab.Ignore
ignore = true
} else {
ignore = false
}
}
}
}

32
vendor/golang.org/x/text/collate/index.go generated vendored
View File

@ -1,32 +0,0 @@
// 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 collate
import "golang.org/x/text/internal/colltab"
const blockSize = 64
func getTable(t tableIndex) *colltab.Table {
return &colltab.Table{
Index: colltab.Trie{
Index0: mainLookup[:][blockSize*t.lookupOffset:],
Values0: mainValues[:][blockSize*t.valuesOffset:],
Index: mainLookup[:],
Values: mainValues[:],
},
ExpandElem: mainExpandElem[:],
ContractTries: colltab.ContractTrieSet(mainCTEntries[:]),
ContractElem: mainContractElem[:],
MaxContractLen: 18,
VariableTop: varTop,
}
}
// tableIndex holds information for constructing a table
// for a certain locale based on the main table.
type tableIndex struct {
lookupOffset uint32
valuesOffset uint32
}

553
vendor/golang.org/x/text/collate/maketables.go generated vendored
View File

@ -1,553 +0,0 @@
// 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.
// +build ignore
// Collation table generator.
// Data read from the web.
package main
import (
"archive/zip"
"bufio"
"bytes"
"flag"
"fmt"
"io"
"io/ioutil"
"log"
"os"
"regexp"
"sort"
"strconv"
"strings"
"unicode/utf8"
"golang.org/x/text/collate"
"golang.org/x/text/collate/build"
"golang.org/x/text/internal/colltab"
"golang.org/x/text/internal/gen"
"golang.org/x/text/language"
"golang.org/x/text/unicode/cldr"
)
var (
test = flag.Bool("test", false,
"test existing tables; can be used to compare web data with package data.")
short = flag.Bool("short", false, `Use "short" alternatives, when available.`)
draft = flag.Bool("draft", false, `Use draft versions, when available.`)
tags = flag.String("tags", "", "build tags to be included after +build directive")
pkg = flag.String("package", "collate",
"the name of the package in which the generated file is to be included")
tables = flagStringSetAllowAll("tables", "collate", "collate,chars",
"comma-spearated list of tables to generate.")
exclude = flagStringSet("exclude", "zh2", "",
"comma-separated list of languages to exclude.")
include = flagStringSet("include", "", "",
"comma-separated list of languages to include. Include trumps exclude.")
// TODO: Not included: unihan gb2312han zhuyin big5han (for size reasons)
// TODO: Not included: traditional (buggy for Bengali)
types = flagStringSetAllowAll("types", "standard,phonebook,phonetic,reformed,pinyin,stroke", "",
"comma-separated list of types that should be included.")
)
// stringSet implements an ordered set based on a list. It implements flag.Value
// to allow a set to be specified as a comma-separated list.
type stringSet struct {
s []string
allowed *stringSet
dirty bool // needs compaction if true
all bool
allowAll bool
}
func flagStringSet(name, def, allowed, usage string) *stringSet {
ss := &stringSet{}
if allowed != "" {
usage += fmt.Sprintf(" (allowed values: any of %s)", allowed)
ss.allowed = &stringSet{}
failOnError(ss.allowed.Set(allowed))
}
ss.Set(def)
flag.Var(ss, name, usage)
return ss
}
func flagStringSetAllowAll(name, def, allowed, usage string) *stringSet {
ss := &stringSet{allowAll: true}
if allowed == "" {
flag.Var(ss, name, usage+fmt.Sprintf(` Use "all" to select all.`))
} else {
ss.allowed = &stringSet{}
failOnError(ss.allowed.Set(allowed))
flag.Var(ss, name, usage+fmt.Sprintf(` (allowed values: "all" or any of %s)`, allowed))
}
ss.Set(def)
return ss
}
func (ss stringSet) Len() int {
return len(ss.s)
}
func (ss stringSet) String() string {
return strings.Join(ss.s, ",")
}
func (ss *stringSet) Set(s string) error {
if ss.allowAll && s == "all" {
ss.s = nil
ss.all = true
return nil
}
ss.s = ss.s[:0]
for _, s := range strings.Split(s, ",") {
if s := strings.TrimSpace(s); s != "" {
if ss.allowed != nil && !ss.allowed.contains(s) {
return fmt.Errorf("unsupported value %q; must be one of %s", s, ss.allowed)
}
ss.add(s)
}
}
ss.compact()
return nil
}
func (ss *stringSet) add(s string) {
ss.s = append(ss.s, s)
ss.dirty = true
}
func (ss *stringSet) values() []string {
ss.compact()
return ss.s
}
func (ss *stringSet) contains(s string) bool {
if ss.all {
return true
}
for _, v := range ss.s {
if v == s {
return true
}
}
return false
}
func (ss *stringSet) compact() {
if !ss.dirty {
return
}
a := ss.s
sort.Strings(a)
k := 0
for i := 1; i < len(a); i++ {
if a[k] != a[i] {
a[k+1] = a[i]
k++
}
}
ss.s = a[:k+1]
ss.dirty = false
}
func skipLang(l string) bool {
if include.Len() > 0 {
return !include.contains(l)
}
return exclude.contains(l)
}
// altInclude returns a list of alternatives (for the LDML alt attribute)
// in order of preference. An empty string in this list indicates the
// default entry.
func altInclude() []string {
l := []string{}
if *short {
l = append(l, "short")
}
l = append(l, "")
// TODO: handle draft using cldr.SetDraftLevel
if *draft {
l = append(l, "proposed")
}
return l
}
func failOnError(e error) {
if e != nil {
log.Panic(e)
}
}
func openArchive() *zip.Reader {
f := gen.OpenCLDRCoreZip()
buffer, err := ioutil.ReadAll(f)
f.Close()
failOnError(err)
archive, err := zip.NewReader(bytes.NewReader(buffer), int64(len(buffer)))
failOnError(err)
return archive
}
// parseUCA parses a Default Unicode Collation Element Table of the format
// specified in https://www.unicode.org/reports/tr10/#File_Format.
// It returns the variable top.
func parseUCA(builder *build.Builder) {
var r io.ReadCloser
var err error
for _, f := range openArchive().File {
if strings.HasSuffix(f.Name, "allkeys_CLDR.txt") {
r, err = f.Open()
}
}
if r == nil {
log.Fatal("File allkeys_CLDR.txt not found in archive.")
}
failOnError(err)
defer r.Close()
scanner := bufio.NewScanner(r)
colelem := regexp.MustCompile(`\[([.*])([0-9A-F.]+)\]`)
for i := 1; scanner.Scan(); i++ {
line := scanner.Text()
if len(line) == 0 || line[0] == '#' {
continue
}
if line[0] == '@' {
// parse properties
switch {
case strings.HasPrefix(line[1:], "version "):
a := strings.Split(line[1:], " ")
if a[1] != gen.UnicodeVersion() {
log.Fatalf("incompatible version %s; want %s", a[1], gen.UnicodeVersion())
}
case strings.HasPrefix(line[1:], "backwards "):
log.Fatalf("%d: unsupported option backwards", i)
default:
log.Printf("%d: unknown option %s", i, line[1:])
}
} else {
// parse entries
part := strings.Split(line, " ; ")
if len(part) != 2 {
log.Fatalf("%d: production rule without ';': %v", i, line)
}
lhs := []rune{}
for _, v := range strings.Split(part[0], " ") {
if v == "" {
continue
}
lhs = append(lhs, rune(convHex(i, v)))
}
var n int
var vars []int
rhs := [][]int{}
for i, m := range colelem.FindAllStringSubmatch(part[1], -1) {
n += len(m[0])
elem := []int{}
for _, h := range strings.Split(m[2], ".") {
elem = append(elem, convHex(i, h))
}
if m[1] == "*" {
vars = append(vars, i)
}
rhs = append(rhs, elem)
}
if len(part[1]) < n+3 || part[1][n+1] != '#' {
log.Fatalf("%d: expected comment; found %s", i, part[1][n:])
}
if *test {
testInput.add(string(lhs))
}
failOnError(builder.Add(lhs, rhs, vars))
}
}
if scanner.Err() != nil {
log.Fatal(scanner.Err())
}
}
func convHex(line int, s string) int {
r, e := strconv.ParseInt(s, 16, 32)
if e != nil {
log.Fatalf("%d: %v", line, e)
}
return int(r)
}
var testInput = stringSet{}
var charRe = regexp.MustCompile(`&#x([0-9A-F]*);`)
var tagRe = regexp.MustCompile(`<([a-z_]*) */>`)
var mainLocales = []string{}
// charsets holds a list of exemplar characters per category.
type charSets map[string][]string
func (p charSets) fprint(w io.Writer) {
fmt.Fprintln(w, "[exN]string{")
for i, k := range []string{"", "contractions", "punctuation", "auxiliary", "currencySymbol", "index"} {
if set := p[k]; len(set) != 0 {
fmt.Fprintf(w, "\t\t%d: %q,\n", i, strings.Join(set, " "))
}
}
fmt.Fprintln(w, "\t},")
}
var localeChars = make(map[string]charSets)
const exemplarHeader = `
type exemplarType int
const (
exCharacters exemplarType = iota
exContractions
exPunctuation
exAuxiliary
exCurrency
exIndex
exN
)
`
func printExemplarCharacters(w io.Writer) {
fmt.Fprintln(w, exemplarHeader)
fmt.Fprintln(w, "var exemplarCharacters = map[string][exN]string{")
for _, loc := range mainLocales {
fmt.Fprintf(w, "\t%q: ", loc)
localeChars[loc].fprint(w)
}
fmt.Fprintln(w, "}")
}
func decodeCLDR(d *cldr.Decoder) *cldr.CLDR {
r := gen.OpenCLDRCoreZip()
data, err := d.DecodeZip(r)
failOnError(err)
return data
}
// parseMain parses XML files in the main directory of the CLDR core.zip file.
func parseMain() {
d := &cldr.Decoder{}
d.SetDirFilter("main")
d.SetSectionFilter("characters")
data := decodeCLDR(d)
for _, loc := range data.Locales() {
x := data.RawLDML(loc)
if skipLang(x.Identity.Language.Type) {
continue
}
if x.Characters != nil {
x, _ = data.LDML(loc)
loc = language.Make(loc).String()
for _, ec := range x.Characters.ExemplarCharacters {
if ec.Draft != "" {
continue
}
if _, ok := localeChars[loc]; !ok {
mainLocales = append(mainLocales, loc)
localeChars[loc] = make(charSets)
}
localeChars[loc][ec.Type] = parseCharacters(ec.Data())
}
}
}
}
func parseCharacters(chars string) []string {
parseSingle := func(s string) (r rune, tail string, escaped bool) {
if s[0] == '\\' {
return rune(s[1]), s[2:], true
}
r, sz := utf8.DecodeRuneInString(s)
return r, s[sz:], false
}
chars = strings.TrimSpace(chars)
if n := len(chars) - 1; chars[n] == ']' && chars[0] == '[' {
chars = chars[1:n]
}
list := []string{}
var r, last, end rune
for len(chars) > 0 {
if chars[0] == '{' { // character sequence
buf := []rune{}
for chars = chars[1:]; len(chars) > 0; {
r, chars, _ = parseSingle(chars)
if r == '}' {
break
}
if r == ' ' {
log.Fatalf("space not supported in sequence %q", chars)
}
buf = append(buf, r)
}
list = append(list, string(buf))
last = 0
} else { // single character
escaped := false
r, chars, escaped = parseSingle(chars)
if r != ' ' {
if r == '-' && !escaped {
if last == 0 {
log.Fatal("'-' should be preceded by a character")
}
end, chars, _ = parseSingle(chars)
for ; last <= end; last++ {
list = append(list, string(last))
}
last = 0
} else {
list = append(list, string(r))
last = r
}
}
}
}
return list
}
var fileRe = regexp.MustCompile(`.*/collation/(.*)\.xml`)
// typeMap translates legacy type keys to their BCP47 equivalent.
var typeMap = map[string]string{
"phonebook": "phonebk",
"traditional": "trad",
}
// parseCollation parses XML files in the collation directory of the CLDR core.zip file.
func parseCollation(b *build.Builder) {
d := &cldr.Decoder{}
d.SetDirFilter("collation")
data := decodeCLDR(d)
for _, loc := range data.Locales() {
x, err := data.LDML(loc)
failOnError(err)
if skipLang(x.Identity.Language.Type) {
continue
}
cs := x.Collations.Collation
sl := cldr.MakeSlice(&cs)
if len(types.s) == 0 {
sl.SelectAnyOf("type", x.Collations.Default())
} else if !types.all {
sl.SelectAnyOf("type", types.s...)
}
sl.SelectOnePerGroup("alt", altInclude())
for _, c := range cs {
id, err := language.Parse(loc)
if err != nil {
fmt.Fprintf(os.Stderr, "invalid locale: %q", err)
continue
}
// Support both old- and new-style defaults.
d := c.Type
if x.Collations.DefaultCollation == nil {
d = x.Collations.Default()
} else {
d = x.Collations.DefaultCollation.Data()
}
// We assume tables are being built either for search or collation,
// but not both. For search the default is always "search".
if d != c.Type && c.Type != "search" {
typ := c.Type
if len(c.Type) > 8 {
typ = typeMap[c.Type]
}
id, err = id.SetTypeForKey("co", typ)
failOnError(err)
}
t := b.Tailoring(id)
c.Process(processor{t})
}
}
}
type processor struct {
t *build.Tailoring
}
func (p processor) Reset(anchor string, before int) (err error) {
if before != 0 {
err = p.t.SetAnchorBefore(anchor)
} else {
err = p.t.SetAnchor(anchor)
}
failOnError(err)
return nil
}
func (p processor) Insert(level int, str, context, extend string) error {
str = context + str
if *test {
testInput.add(str)
}
// TODO: mimic bug in old maketables: remove.
err := p.t.Insert(colltab.Level(level-1), str, context+extend)
failOnError(err)
return nil
}
func (p processor) Index(id string) {
}
func testCollator(c *collate.Collator) {
c0 := collate.New(language.Und)
// iterator over all characters for all locales and check
// whether Key is equal.
buf := collate.Buffer{}
// Add all common and not too uncommon runes to the test set.
for i := rune(0); i < 0x30000; i++ {
testInput.add(string(i))
}
for i := rune(0xE0000); i < 0xF0000; i++ {
testInput.add(string(i))
}
for _, str := range testInput.values() {
k0 := c0.KeyFromString(&buf, str)
k := c.KeyFromString(&buf, str)
if !bytes.Equal(k0, k) {
failOnError(fmt.Errorf("test:%U: keys differ (%x vs %x)", []rune(str), k0, k))
}
buf.Reset()
}
fmt.Println("PASS")
}
func main() {
gen.Init()
b := build.NewBuilder()
parseUCA(b)
if tables.contains("chars") {
parseMain()
}
parseCollation(b)
c, err := b.Build()
failOnError(err)
if *test {
testCollator(collate.NewFromTable(c))
} else {
w := &bytes.Buffer{}
gen.WriteUnicodeVersion(w)
gen.WriteCLDRVersion(w)
if tables.contains("collate") {
_, err = b.Print(w)
failOnError(err)
}
if tables.contains("chars") {
printExemplarCharacters(w)
}
gen.WriteGoFile("tables.go", *pkg, w.Bytes())
}
}

239
vendor/golang.org/x/text/collate/option.go generated vendored
View File

@ -1,239 +0,0 @@
// Copyright 2014 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 collate
import (
"sort"
"golang.org/x/text/internal/colltab"
"golang.org/x/text/language"
"golang.org/x/text/unicode/norm"
)
// newCollator creates a new collator with default options configured.
func newCollator(t colltab.Weighter) *Collator {
// Initialize a collator with default options.
c := &Collator{
options: options{
ignore: [colltab.NumLevels]bool{
colltab.Quaternary: true,
colltab.Identity: true,
},
f: norm.NFD,
t: t,
},
}
// TODO: store vt in tags or remove.
c.variableTop = t.Top()
return c
}
// An Option is used to change the behavior of a Collator. Options override the
// settings passed through the locale identifier.
type Option struct {
priority int
f func(o *options)
}
type prioritizedOptions []Option
func (p prioritizedOptions) Len() int {
return len(p)
}
func (p prioritizedOptions) Swap(i, j int) {
p[i], p[j] = p[j], p[i]
}
func (p prioritizedOptions) Less(i, j int) bool {
return p[i].priority < p[j].priority
}
type options struct {
// ignore specifies which levels to ignore.
ignore [colltab.NumLevels]bool
// caseLevel is true if there is an additional level of case matching
// between the secondary and tertiary levels.
caseLevel bool
// backwards specifies the order of sorting at the secondary level.
// This option exists predominantly to support reverse sorting of accents in French.
backwards bool
// numeric specifies whether any sequence of decimal digits (category is Nd)
// is sorted at a primary level with its numeric value.
// For example, "A-21" < "A-123".
// This option is set by wrapping the main Weighter with NewNumericWeighter.
numeric bool
// alternate specifies an alternative handling of variables.
alternate alternateHandling
// variableTop is the largest primary value that is considered to be
// variable.
variableTop uint32
t colltab.Weighter
f norm.Form
}
func (o *options) setOptions(opts []Option) {
sort.Sort(prioritizedOptions(opts))
for _, x := range opts {
x.f(o)
}
}
// OptionsFromTag extracts the BCP47 collation options from the tag and
// configures a collator accordingly. These options are set before any other
// option.
func OptionsFromTag(t language.Tag) Option {
return Option{0, func(o *options) {
o.setFromTag(t)
}}
}
func (o *options) setFromTag(t language.Tag) {
o.caseLevel = ldmlBool(t, o.caseLevel, "kc")
o.backwards = ldmlBool(t, o.backwards, "kb")
o.numeric = ldmlBool(t, o.numeric, "kn")
// Extract settings from the BCP47 u extension.
switch t.TypeForKey("ks") { // strength
case "level1":
o.ignore[colltab.Secondary] = true
o.ignore[colltab.Tertiary] = true
case "level2":
o.ignore[colltab.Tertiary] = true
case "level3", "":
// The default.
case "level4":
o.ignore[colltab.Quaternary] = false
case "identic":
o.ignore[colltab.Quaternary] = false
o.ignore[colltab.Identity] = false
}
switch t.TypeForKey("ka") {
case "shifted":
o.alternate = altShifted
// The following two types are not official BCP47, but we support them to
// give access to this otherwise hidden functionality. The name blanked is
// derived from the LDML name blanked and posix reflects the main use of
// the shift-trimmed option.
case "blanked":
o.alternate = altBlanked
case "posix":
o.alternate = altShiftTrimmed
}
// TODO: caseFirst ("kf"), reorder ("kr"), and maybe variableTop ("vt").
// Not used:
// - normalization ("kk", not necessary for this implementation)
// - hiraganaQuatenary ("kh", obsolete)
}
func ldmlBool(t language.Tag, old bool, key string) bool {
switch t.TypeForKey(key) {
case "true":
return true
case "false":
return false
default:
return old
}
}
var (
// IgnoreCase sets case-insensitive comparison.
IgnoreCase Option = ignoreCase
ignoreCase = Option{3, ignoreCaseF}
// IgnoreDiacritics causes diacritical marks to be ignored. ("o" == "ö").
IgnoreDiacritics Option = ignoreDiacritics
ignoreDiacritics = Option{3, ignoreDiacriticsF}
// IgnoreWidth causes full-width characters to match their half-width
// equivalents.
IgnoreWidth Option = ignoreWidth
ignoreWidth = Option{2, ignoreWidthF}
// Loose sets the collator to ignore diacritics, case and width.
Loose Option = loose
loose = Option{4, looseF}
// Force ordering if strings are equivalent but not equal.
Force Option = force
force = Option{5, forceF}
// Numeric specifies that numbers should sort numerically ("2" < "12").
Numeric Option = numeric
numeric = Option{5, numericF}
)
func ignoreWidthF(o *options) {
o.ignore[colltab.Tertiary] = true
o.caseLevel = true
}
func ignoreDiacriticsF(o *options) {
o.ignore[colltab.Secondary] = true
}
func ignoreCaseF(o *options) {
o.ignore[colltab.Tertiary] = true
o.caseLevel = false
}
func looseF(o *options) {
ignoreWidthF(o)
ignoreDiacriticsF(o)
ignoreCaseF(o)
}
func forceF(o *options) {
o.ignore[colltab.Identity] = false
}
func numericF(o *options) { o.numeric = true }
// Reorder overrides the pre-defined ordering of scripts and character sets.
func Reorder(s ...string) Option {
// TODO: need fractional weights to implement this.
panic("TODO: implement")
}
// TODO: consider making these public again. These options cannot be fully
// specified in BCP47, so an API interface seems warranted. Still a higher-level
// interface would be nice (e.g. a POSIX option for enabling altShiftTrimmed)
// alternateHandling identifies the various ways in which variables are handled.
// A rune with a primary weight lower than the variable top is considered a
// variable.
// See https://www.unicode.org/reports/tr10/#Variable_Weighting for details.
type alternateHandling int
const (
// altNonIgnorable turns off special handling of variables.
altNonIgnorable alternateHandling = iota
// altBlanked sets variables and all subsequent primary ignorables to be
// ignorable at all levels. This is identical to removing all variables
// and subsequent primary ignorables from the input.
altBlanked
// altShifted sets variables to be ignorable for levels one through three and
// adds a fourth level based on the values of the ignored levels.
altShifted
// altShiftTrimmed is a slight variant of altShifted that is used to
// emulate POSIX.
altShiftTrimmed
)

81
vendor/golang.org/x/text/collate/sort.go generated vendored
View File

@ -1,81 +0,0 @@
// 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 collate
import (
"bytes"
"sort"
)
const (
maxSortBuffer = 40960
maxSortEntries = 4096
)
type swapper interface {
Swap(i, j int)
}
type sorter struct {
buf *Buffer
keys [][]byte
src swapper
}
func (s *sorter) init(n int) {
if s.buf == nil {
s.buf = &Buffer{}
s.buf.init()
}
if cap(s.keys) < n {
s.keys = make([][]byte, n)
}
s.keys = s.keys[0:n]
}
func (s *sorter) sort(src swapper) {
s.src = src
sort.Sort(s)
}
func (s sorter) Len() int {
return len(s.keys)
}
func (s sorter) Less(i, j int) bool {
return bytes.Compare(s.keys[i], s.keys[j]) == -1
}
func (s sorter) Swap(i, j int) {
s.keys[i], s.keys[j] = s.keys[j], s.keys[i]
s.src.Swap(i, j)
}
// A Lister can be sorted by Collator's Sort method.
type Lister interface {
Len() int
Swap(i, j int)
// Bytes returns the bytes of the text at index i.
Bytes(i int) []byte
}
// Sort uses sort.Sort to sort the strings represented by x using the rules of c.
func (c *Collator) Sort(x Lister) {
n := x.Len()
c.sorter.init(n)
for i := 0; i < n; i++ {
c.sorter.keys[i] = c.Key(c.sorter.buf, x.Bytes(i))
}
c.sorter.sort(x)
}
// SortStrings uses sort.Sort to sort the strings in x using the rules of c.
func (c *Collator) SortStrings(x []string) {
c.sorter.init(len(x))
for i, s := range x {
c.sorter.keys[i] = c.KeyFromString(c.sorter.buf, s)
}
c.sorter.sort(sort.StringSlice(x))
}

73789
vendor/golang.org/x/text/collate/tables.go generated vendored
View File

File diff suppressed because it is too large Load Diff

556
vendor/golang.org/x/text/encoding/charmap/maketables.go generated vendored
View File

@ -1,556 +0,0 @@
// 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.
// +build ignore
package main
import (
"bufio"
"fmt"
"log"
"net/http"
"sort"
"strings"
"unicode/utf8"
"golang.org/x/text/encoding"
"golang.org/x/text/internal/gen"
)
const ascii = "\x00\x01\x02\x03\x04\x05\x06\x07\x08\x09\x0a\x0b\x0c\x0d\x0e\x0f" +
"\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f" +
` !"#$%&'()*+,-./0123456789:;<=>?` +
`@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_` +
"`abcdefghijklmnopqrstuvwxyz{|}~\u007f"
var encodings = []struct {
name string
mib string
comment string
varName string
replacement byte
mapping string
}{
{
"IBM Code Page 037",
"IBM037",
"",
"CodePage037",
0x3f,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM037-2.1.2.ucm",
},
{
"IBM Code Page 437",
"PC8CodePage437",
"",
"CodePage437",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM437-2.1.2.ucm",
},
{
"IBM Code Page 850",
"PC850Multilingual",
"",
"CodePage850",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM850-2.1.2.ucm",
},
{
"IBM Code Page 852",
"PCp852",
"",
"CodePage852",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM852-2.1.2.ucm",
},
{
"IBM Code Page 855",
"IBM855",
"",
"CodePage855",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM855-2.1.2.ucm",
},
{
"Windows Code Page 858", // PC latin1 with Euro
"IBM00858",
"",
"CodePage858",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/windows-858-2000.ucm",
},
{
"IBM Code Page 860",
"IBM860",
"",
"CodePage860",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM860-2.1.2.ucm",
},
{
"IBM Code Page 862",
"PC862LatinHebrew",
"",
"CodePage862",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM862-2.1.2.ucm",
},
{
"IBM Code Page 863",
"IBM863",
"",
"CodePage863",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM863-2.1.2.ucm",
},
{
"IBM Code Page 865",
"IBM865",
"",
"CodePage865",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM865-2.1.2.ucm",
},
{
"IBM Code Page 866",
"IBM866",
"",
"CodePage866",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-ibm866.txt",
},
{
"IBM Code Page 1047",
"IBM1047",
"",
"CodePage1047",
0x3f,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/glibc-IBM1047-2.1.2.ucm",
},
{
"IBM Code Page 1140",
"IBM01140",
"",
"CodePage1140",
0x3f,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/ibm-1140_P100-1997.ucm",
},
{
"ISO 8859-1",
"ISOLatin1",
"",
"ISO8859_1",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/iso-8859_1-1998.ucm",
},
{
"ISO 8859-2",
"ISOLatin2",
"",
"ISO8859_2",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-2.txt",
},
{
"ISO 8859-3",
"ISOLatin3",
"",
"ISO8859_3",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-3.txt",
},
{
"ISO 8859-4",
"ISOLatin4",
"",
"ISO8859_4",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-4.txt",
},
{
"ISO 8859-5",
"ISOLatinCyrillic",
"",
"ISO8859_5",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-5.txt",
},
{
"ISO 8859-6",
"ISOLatinArabic",
"",
"ISO8859_6,ISO8859_6E,ISO8859_6I",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-6.txt",
},
{
"ISO 8859-7",
"ISOLatinGreek",
"",
"ISO8859_7",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-7.txt",
},
{
"ISO 8859-8",
"ISOLatinHebrew",
"",
"ISO8859_8,ISO8859_8E,ISO8859_8I",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-8.txt",
},
{
"ISO 8859-9",
"ISOLatin5",
"",
"ISO8859_9",
encoding.ASCIISub,
"http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/iso-8859_9-1999.ucm",
},
{
"ISO 8859-10",
"ISOLatin6",
"",
"ISO8859_10",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-10.txt",
},
{
"ISO 8859-13",
"ISO885913",
"",
"ISO8859_13",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-13.txt",
},
{
"ISO 8859-14",
"ISO885914",
"",
"ISO8859_14",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-14.txt",
},
{
"ISO 8859-15",
"ISO885915",
"",
"ISO8859_15",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-15.txt",
},
{
"ISO 8859-16",
"ISO885916",
"",
"ISO8859_16",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-iso-8859-16.txt",
},
{
"KOI8-R",
"KOI8R",
"",
"KOI8R",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-koi8-r.txt",
},
{
"KOI8-U",
"KOI8U",
"",
"KOI8U",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-koi8-u.txt",
},
{
"Macintosh",
"Macintosh",
"",
"Macintosh",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-macintosh.txt",
},
{
"Macintosh Cyrillic",
"MacintoshCyrillic",
"",
"MacintoshCyrillic",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-x-mac-cyrillic.txt",
},
{
"Windows 874",
"Windows874",
"",
"Windows874",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-874.txt",
},
{
"Windows 1250",
"Windows1250",
"",
"Windows1250",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1250.txt",
},
{
"Windows 1251",
"Windows1251",
"",
"Windows1251",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1251.txt",
},
{
"Windows 1252",
"Windows1252",
"",
"Windows1252",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1252.txt",
},
{
"Windows 1253",
"Windows1253",
"",
"Windows1253",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1253.txt",
},
{
"Windows 1254",
"Windows1254",
"",
"Windows1254",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1254.txt",
},
{
"Windows 1255",
"Windows1255",
"",
"Windows1255",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1255.txt",
},
{
"Windows 1256",
"Windows1256",
"",
"Windows1256",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1256.txt",
},
{
"Windows 1257",
"Windows1257",
"",
"Windows1257",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1257.txt",
},
{
"Windows 1258",
"Windows1258",
"",
"Windows1258",
encoding.ASCIISub,
"http://encoding.spec.whatwg.org/index-windows-1258.txt",
},
{
"X-User-Defined",
"XUserDefined",
"It is defined at http://encoding.spec.whatwg.org/#x-user-defined",
"XUserDefined",
encoding.ASCIISub,
ascii +
"\uf780\uf781\uf782\uf783\uf784\uf785\uf786\uf787" +
"\uf788\uf789\uf78a\uf78b\uf78c\uf78d\uf78e\uf78f" +
"\uf790\uf791\uf792\uf793\uf794\uf795\uf796\uf797" +
"\uf798\uf799\uf79a\uf79b\uf79c\uf79d\uf79e\uf79f" +
"\uf7a0\uf7a1\uf7a2\uf7a3\uf7a4\uf7a5\uf7a6\uf7a7" +
"\uf7a8\uf7a9\uf7aa\uf7ab\uf7ac\uf7ad\uf7ae\uf7af" +
"\uf7b0\uf7b1\uf7b2\uf7b3\uf7b4\uf7b5\uf7b6\uf7b7" +
"\uf7b8\uf7b9\uf7ba\uf7bb\uf7bc\uf7bd\uf7be\uf7bf" +
"\uf7c0\uf7c1\uf7c2\uf7c3\uf7c4\uf7c5\uf7c6\uf7c7" +
"\uf7c8\uf7c9\uf7ca\uf7cb\uf7cc\uf7cd\uf7ce\uf7cf" +
"\uf7d0\uf7d1\uf7d2\uf7d3\uf7d4\uf7d5\uf7d6\uf7d7" +
"\uf7d8\uf7d9\uf7da\uf7db\uf7dc\uf7dd\uf7de\uf7df" +
"\uf7e0\uf7e1\uf7e2\uf7e3\uf7e4\uf7e5\uf7e6\uf7e7" +
"\uf7e8\uf7e9\uf7ea\uf7eb\uf7ec\uf7ed\uf7ee\uf7ef" +
"\uf7f0\uf7f1\uf7f2\uf7f3\uf7f4\uf7f5\uf7f6\uf7f7" +
"\uf7f8\uf7f9\uf7fa\uf7fb\uf7fc\uf7fd\uf7fe\uf7ff",
},
}
func getWHATWG(url string) string {
res, err := http.Get(url)
if err != nil {
log.Fatalf("%q: Get: %v", url, err)
}
defer res.Body.Close()
mapping := make([]rune, 128)
for i := range mapping {
mapping[i] = '\ufffd'
}
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := 0, 0
if _, err := fmt.Sscanf(s, "%d\t0x%x", &x, &y); err != nil {
log.Fatalf("could not parse %q", s)
}
if x < 0 || 128 <= x {
log.Fatalf("code %d is out of range", x)
}
if 0x80 <= y && y < 0xa0 {
// We diverge from the WHATWG spec by mapping control characters
// in the range [0x80, 0xa0) to U+FFFD.
continue
}
mapping[x] = rune(y)
}
return ascii + string(mapping)
}
func getUCM(url string) string {
res, err := http.Get(url)
if err != nil {
log.Fatalf("%q: Get: %v", url, err)
}
defer res.Body.Close()
mapping := make([]rune, 256)
for i := range mapping {
mapping[i] = '\ufffd'
}
charsFound := 0
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
var c byte
var r rune
if _, err := fmt.Sscanf(s, `<U%x> \x%x |0`, &r, &c); err != nil {
continue
}
mapping[c] = r
charsFound++
}
if charsFound < 200 {
log.Fatalf("%q: only %d characters found (wrong page format?)", url, charsFound)
}
return string(mapping)
}
func main() {
mibs := map[string]bool{}
all := []string{}
w := gen.NewCodeWriter()
defer w.WriteGoFile("tables.go", "charmap")
printf := func(s string, a ...interface{}) { fmt.Fprintf(w, s, a...) }
printf("import (\n")
printf("\t\"golang.org/x/text/encoding\"\n")
printf("\t\"golang.org/x/text/encoding/internal/identifier\"\n")
printf(")\n\n")
for _, e := range encodings {
varNames := strings.Split(e.varName, ",")
all = append(all, varNames...)
varName := varNames[0]
switch {
case strings.HasPrefix(e.mapping, "http://encoding.spec.whatwg.org/"):
e.mapping = getWHATWG(e.mapping)
case strings.HasPrefix(e.mapping, "http://source.icu-project.org/repos/icu/data/trunk/charset/data/ucm/"):
e.mapping = getUCM(e.mapping)
}
asciiSuperset, low := strings.HasPrefix(e.mapping, ascii), 0x00
if asciiSuperset {
low = 0x80
}
lvn := 1
if strings.HasPrefix(varName, "ISO") || strings.HasPrefix(varName, "KOI") {
lvn = 3
}
lowerVarName := strings.ToLower(varName[:lvn]) + varName[lvn:]
printf("// %s is the %s encoding.\n", varName, e.name)
if e.comment != "" {
printf("//\n// %s\n", e.comment)
}
printf("var %s *Charmap = &%s\n\nvar %s = Charmap{\nname: %q,\n",
varName, lowerVarName, lowerVarName, e.name)
if mibs[e.mib] {
log.Fatalf("MIB type %q declared multiple times.", e.mib)
}
printf("mib: identifier.%s,\n", e.mib)
printf("asciiSuperset: %t,\n", asciiSuperset)
printf("low: 0x%02x,\n", low)
printf("replacement: 0x%02x,\n", e.replacement)
printf("decode: [256]utf8Enc{\n")
i, backMapping := 0, map[rune]byte{}
for _, c := range e.mapping {
if _, ok := backMapping[c]; !ok && c != utf8.RuneError {
backMapping[c] = byte(i)
}
var buf [8]byte
n := utf8.EncodeRune(buf[:], c)
if n > 3 {
panic(fmt.Sprintf("rune %q (%U) is too long", c, c))
}
printf("{%d,[3]byte{0x%02x,0x%02x,0x%02x}},", n, buf[0], buf[1], buf[2])
if i%2 == 1 {
printf("\n")
}
i++
}
printf("},\n")
printf("encode: [256]uint32{\n")
encode := make([]uint32, 0, 256)
for c, i := range backMapping {
encode = append(encode, uint32(i)<<24|uint32(c))
}
sort.Sort(byRune(encode))
for len(encode) < cap(encode) {
encode = append(encode, encode[len(encode)-1])
}
for i, enc := range encode {
printf("0x%08x,", enc)
if i%8 == 7 {
printf("\n")
}
}
printf("},\n}\n")
// Add an estimate of the size of a single Charmap{} struct value, which
// includes two 256 elem arrays of 4 bytes and some extra fields, which
// align to 3 uint64s on 64-bit architectures.
w.Size += 2*4*256 + 3*8
}
// TODO: add proper line breaking.
printf("var listAll = []encoding.Encoding{\n%s,\n}\n\n", strings.Join(all, ",\n"))
}
type byRune []uint32
func (b byRune) Len() int { return len(b) }
func (b byRune) Less(i, j int) bool { return b[i]&0xffffff < b[j]&0xffffff }
func (b byRune) Swap(i, j int) { b[i], b[j] = b[j], b[i] }

173
vendor/golang.org/x/text/encoding/htmlindex/gen.go generated vendored
View File

@ -1,173 +0,0 @@
// Copyright 2015 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
package main
import (
"bytes"
"encoding/json"
"fmt"
"log"
"strings"
"golang.org/x/text/internal/gen"
)
type group struct {
Encodings []struct {
Labels []string
Name string
}
}
func main() {
gen.Init()
r := gen.Open("https://encoding.spec.whatwg.org", "whatwg", "encodings.json")
var groups []group
if err := json.NewDecoder(r).Decode(&groups); err != nil {
log.Fatalf("Error reading encodings.json: %v", err)
}
w := &bytes.Buffer{}
fmt.Fprintln(w, "type htmlEncoding byte")
fmt.Fprintln(w, "const (")
for i, g := range groups {
for _, e := range g.Encodings {
key := strings.ToLower(e.Name)
name := consts[key]
if name == "" {
log.Fatalf("No const defined for %s.", key)
}
if i == 0 {
fmt.Fprintf(w, "%s htmlEncoding = iota\n", name)
} else {
fmt.Fprintf(w, "%s\n", name)
}
}
}
fmt.Fprintln(w, "numEncodings")
fmt.Fprint(w, ")\n\n")
fmt.Fprintln(w, "var canonical = [numEncodings]string{")
for _, g := range groups {
for _, e := range g.Encodings {
fmt.Fprintf(w, "%q,\n", strings.ToLower(e.Name))
}
}
fmt.Fprint(w, "}\n\n")
fmt.Fprintln(w, "var nameMap = map[string]htmlEncoding{")
for _, g := range groups {
for _, e := range g.Encodings {
for _, l := range e.Labels {
key := strings.ToLower(e.Name)
name := consts[key]
fmt.Fprintf(w, "%q: %s,\n", l, name)
}
}
}
fmt.Fprint(w, "}\n\n")
var tags []string
fmt.Fprintln(w, "var localeMap = []htmlEncoding{")
for _, loc := range locales {
tags = append(tags, loc.tag)
fmt.Fprintf(w, "%s, // %s \n", consts[loc.name], loc.tag)
}
fmt.Fprint(w, "}\n\n")
fmt.Fprintf(w, "const locales = %q\n", strings.Join(tags, " "))
gen.WriteGoFile("tables.go", "htmlindex", w.Bytes())
}
// consts maps canonical encoding name to internal constant.
var consts = map[string]string{
"utf-8": "utf8",
"ibm866": "ibm866",
"iso-8859-2": "iso8859_2",
"iso-8859-3": "iso8859_3",
"iso-8859-4": "iso8859_4",
"iso-8859-5": "iso8859_5",
"iso-8859-6": "iso8859_6",
"iso-8859-7": "iso8859_7",
"iso-8859-8": "iso8859_8",
"iso-8859-8-i": "iso8859_8I",
"iso-8859-10": "iso8859_10",
"iso-8859-13": "iso8859_13",
"iso-8859-14": "iso8859_14",
"iso-8859-15": "iso8859_15",
"iso-8859-16": "iso8859_16",
"koi8-r": "koi8r",
"koi8-u": "koi8u",
"macintosh": "macintosh",
"windows-874": "windows874",
"windows-1250": "windows1250",
"windows-1251": "windows1251",
"windows-1252": "windows1252",
"windows-1253": "windows1253",
"windows-1254": "windows1254",
"windows-1255": "windows1255",
"windows-1256": "windows1256",
"windows-1257": "windows1257",
"windows-1258": "windows1258",
"x-mac-cyrillic": "macintoshCyrillic",
"gbk": "gbk",
"gb18030": "gb18030",
// "hz-gb-2312": "hzgb2312", // Was removed from WhatWG
"big5": "big5",
"euc-jp": "eucjp",
"iso-2022-jp": "iso2022jp",
"shift_jis": "shiftJIS",
"euc-kr": "euckr",
"replacement": "replacement",
"utf-16be": "utf16be",
"utf-16le": "utf16le",
"x-user-defined": "xUserDefined",
}
// locales is taken from
// https://html.spec.whatwg.org/multipage/syntax.html#encoding-sniffing-algorithm.
var locales = []struct{ tag, name string }{
// The default value. Explicitly state latin to benefit from the exact
// script option, while still making 1252 the default encoding for languages
// written in Latin script.
{"und_Latn", "windows-1252"},
{"ar", "windows-1256"},
{"ba", "windows-1251"},
{"be", "windows-1251"},
{"bg", "windows-1251"},
{"cs", "windows-1250"},
{"el", "iso-8859-7"},
{"et", "windows-1257"},
{"fa", "windows-1256"},
{"he", "windows-1255"},
{"hr", "windows-1250"},
{"hu", "iso-8859-2"},
{"ja", "shift_jis"},
{"kk", "windows-1251"},
{"ko", "euc-kr"},
{"ku", "windows-1254"},
{"ky", "windows-1251"},
{"lt", "windows-1257"},
{"lv", "windows-1257"},
{"mk", "windows-1251"},
{"pl", "iso-8859-2"},
{"ru", "windows-1251"},
{"sah", "windows-1251"},
{"sk", "windows-1250"},
{"sl", "iso-8859-2"},
{"sr", "windows-1251"},
{"tg", "windows-1251"},
{"th", "windows-874"},
{"tr", "windows-1254"},
{"tt", "windows-1251"},
{"uk", "windows-1251"},
{"vi", "windows-1258"},
{"zh-hans", "gb18030"},
{"zh-hant", "big5"},
}

142
vendor/golang.org/x/text/encoding/internal/identifier/gen.go generated vendored
View File

@ -1,142 +0,0 @@
// Copyright 2015 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
package main
import (
"bytes"
"encoding/xml"
"fmt"
"io"
"log"
"strings"
"golang.org/x/text/internal/gen"
)
type registry struct {
XMLName xml.Name `xml:"registry"`
Updated string `xml:"updated"`
Registry []struct {
ID string `xml:"id,attr"`
Record []struct {
Name string `xml:"name"`
Xref []struct {
Type string `xml:"type,attr"`
Data string `xml:"data,attr"`
} `xml:"xref"`
Desc struct {
Data string `xml:",innerxml"`
// Any []struct {
// Data string `xml:",chardata"`
// } `xml:",any"`
// Data string `xml:",chardata"`
} `xml:"description,"`
MIB string `xml:"value"`
Alias []string `xml:"alias"`
MIME string `xml:"preferred_alias"`
} `xml:"record"`
} `xml:"registry"`
}
func main() {
r := gen.OpenIANAFile("assignments/character-sets/character-sets.xml")
reg := &registry{}
if err := xml.NewDecoder(r).Decode(&reg); err != nil && err != io.EOF {
log.Fatalf("Error decoding charset registry: %v", err)
}
if len(reg.Registry) == 0 || reg.Registry[0].ID != "character-sets-1" {
log.Fatalf("Unexpected ID %s", reg.Registry[0].ID)
}
w := &bytes.Buffer{}
fmt.Fprintf(w, "const (\n")
for _, rec := range reg.Registry[0].Record {
constName := ""
for _, a := range rec.Alias {
if strings.HasPrefix(a, "cs") && strings.IndexByte(a, '-') == -1 {
// Some of the constant definitions have comments in them. Strip those.
constName = strings.Title(strings.SplitN(a[2:], "\n", 2)[0])
}
}
if constName == "" {
switch rec.MIB {
case "2085":
constName = "HZGB2312" // Not listed as alias for some reason.
default:
log.Fatalf("No cs alias defined for %s.", rec.MIB)
}
}
if rec.MIME != "" {
rec.MIME = fmt.Sprintf(" (MIME: %s)", rec.MIME)
}
fmt.Fprintf(w, "// %s is the MIB identifier with IANA name %s%s.\n//\n", constName, rec.Name, rec.MIME)
if len(rec.Desc.Data) > 0 {
fmt.Fprint(w, "// ")
d := xml.NewDecoder(strings.NewReader(rec.Desc.Data))
inElem := true
attr := ""
for {
t, err := d.Token()
if err != nil {
if err != io.EOF {
log.Fatal(err)
}
break
}
switch x := t.(type) {
case xml.CharData:
attr = "" // Don't need attribute info.
a := bytes.Split([]byte(x), []byte("\n"))
for i, b := range a {
if b = bytes.TrimSpace(b); len(b) != 0 {
if !inElem && i > 0 {
fmt.Fprint(w, "\n// ")
}
inElem = false
fmt.Fprintf(w, "%s ", string(b))
}
}
case xml.StartElement:
if x.Name.Local == "xref" {
inElem = true
use := false
for _, a := range x.Attr {
if a.Name.Local == "type" {
use = use || a.Value != "person"
}
if a.Name.Local == "data" && use {
// Patch up URLs to use https. From some links, the
// https version is different from the http one.
s := a.Value
s = strings.Replace(s, "http://", "https://", -1)
s = strings.Replace(s, "/unicode/", "/", -1)
attr = s + " "
}
}
}
case xml.EndElement:
inElem = false
fmt.Fprint(w, attr)
}
}
fmt.Fprint(w, "\n")
}
for _, x := range rec.Xref {
switch x.Type {
case "rfc":
fmt.Fprintf(w, "// Reference: %s\n", strings.ToUpper(x.Data))
case "uri":
fmt.Fprintf(w, "// Reference: %s\n", x.Data)
}
}
fmt.Fprintf(w, "%s MIB = %s\n", constName, rec.MIB)
fmt.Fprintln(w)
}
fmt.Fprintln(w, ")")
gen.WriteGoFile("mib.go", "identifier", w.Bytes())
}

161
vendor/golang.org/x/text/encoding/japanese/maketables.go generated vendored
View File

@ -1,161 +0,0 @@
// 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.
// +build ignore
package main
// This program generates tables.go:
// go run maketables.go | gofmt > tables.go
// TODO: Emoji extensions?
// https://www.unicode.org/faq/emoji_dingbats.html
// https://www.unicode.org/Public/UNIDATA/EmojiSources.txt
import (
"bufio"
"fmt"
"log"
"net/http"
"sort"
"strings"
)
type entry struct {
jisCode, table int
}
func main() {
fmt.Printf("// generated by go run maketables.go; DO NOT EDIT\n\n")
fmt.Printf("// Package japanese provides Japanese encodings such as EUC-JP and Shift JIS.\n")
fmt.Printf(`package japanese // import "golang.org/x/text/encoding/japanese"` + "\n\n")
reverse := [65536]entry{}
for i := range reverse {
reverse[i].table = -1
}
tables := []struct {
url string
name string
}{
{"http://encoding.spec.whatwg.org/index-jis0208.txt", "0208"},
{"http://encoding.spec.whatwg.org/index-jis0212.txt", "0212"},
}
for i, table := range tables {
res, err := http.Get(table.url)
if err != nil {
log.Fatalf("%q: Get: %v", table.url, err)
}
defer res.Body.Close()
mapping := [65536]uint16{}
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := 0, uint16(0)
if _, err := fmt.Sscanf(s, "%d 0x%x", &x, &y); err != nil {
log.Fatalf("%q: could not parse %q", table.url, s)
}
if x < 0 || 120*94 <= x {
log.Fatalf("%q: JIS code %d is out of range", table.url, x)
}
mapping[x] = y
if reverse[y].table == -1 {
reverse[y] = entry{jisCode: x, table: i}
}
}
if err := scanner.Err(); err != nil {
log.Fatalf("%q: scanner error: %v", table.url, err)
}
fmt.Printf("// jis%sDecode is the decoding table from JIS %s code to Unicode.\n// It is defined at %s\n",
table.name, table.name, table.url)
fmt.Printf("var jis%sDecode = [...]uint16{\n", table.name)
for i, m := range mapping {
if m != 0 {
fmt.Printf("\t%d: 0x%04X,\n", i, m)
}
}
fmt.Printf("}\n\n")
}
// Any run of at least separation continuous zero entries in the reverse map will
// be a separate encode table.
const separation = 1024
intervals := []interval(nil)
low, high := -1, -1
for i, v := range reverse {
if v.table == -1 {
continue
}
if low < 0 {
low = i
} else if i-high >= separation {
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
low = i
}
high = i + 1
}
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
sort.Sort(byDecreasingLength(intervals))
fmt.Printf("const (\n")
fmt.Printf("\tjis0208 = 1\n")
fmt.Printf("\tjis0212 = 2\n")
fmt.Printf("\tcodeMask = 0x7f\n")
fmt.Printf("\tcodeShift = 7\n")
fmt.Printf("\ttableShift = 14\n")
fmt.Printf(")\n\n")
fmt.Printf("const numEncodeTables = %d\n\n", len(intervals))
fmt.Printf("// encodeX are the encoding tables from Unicode to JIS code,\n")
fmt.Printf("// sorted by decreasing length.\n")
for i, v := range intervals {
fmt.Printf("// encode%d: %5d entries for runes in [%5d, %5d).\n", i, v.len(), v.low, v.high)
}
fmt.Printf("//\n")
fmt.Printf("// The high two bits of the value record whether the JIS code comes from the\n")
fmt.Printf("// JIS0208 table (high bits == 1) or the JIS0212 table (high bits == 2).\n")
fmt.Printf("// The low 14 bits are two 7-bit unsigned integers j1 and j2 that form the\n")
fmt.Printf("// JIS code (94*j1 + j2) within that table.\n")
fmt.Printf("\n")
for i, v := range intervals {
fmt.Printf("const encode%dLow, encode%dHigh = %d, %d\n\n", i, i, v.low, v.high)
fmt.Printf("var encode%d = [...]uint16{\n", i)
for j := v.low; j < v.high; j++ {
x := reverse[j]
if x.table == -1 {
continue
}
fmt.Printf("\t%d - %d: jis%s<<14 | 0x%02X<<7 | 0x%02X,\n",
j, v.low, tables[x.table].name, x.jisCode/94, x.jisCode%94)
}
fmt.Printf("}\n\n")
}
}
// interval is a half-open interval [low, high).
type interval struct {
low, high int
}
func (i interval) len() int { return i.high - i.low }
// byDecreasingLength sorts intervals by decreasing length.
type byDecreasingLength []interval
func (b byDecreasingLength) Len() int { return len(b) }
func (b byDecreasingLength) Less(i, j int) bool { return b[i].len() > b[j].len() }
func (b byDecreasingLength) Swap(i, j int) { b[i], b[j] = b[j], b[i] }

143
vendor/golang.org/x/text/encoding/korean/maketables.go generated vendored
View File

@ -1,143 +0,0 @@
// 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.
// +build ignore
package main
// This program generates tables.go:
// go run maketables.go | gofmt > tables.go
import (
"bufio"
"fmt"
"log"
"net/http"
"sort"
"strings"
)
func main() {
fmt.Printf("// generated by go run maketables.go; DO NOT EDIT\n\n")
fmt.Printf("// Package korean provides Korean encodings such as EUC-KR.\n")
fmt.Printf(`package korean // import "golang.org/x/text/encoding/korean"` + "\n\n")
res, err := http.Get("http://encoding.spec.whatwg.org/index-euc-kr.txt")
if err != nil {
log.Fatalf("Get: %v", err)
}
defer res.Body.Close()
mapping := [65536]uint16{}
reverse := [65536]uint16{}
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := uint16(0), uint16(0)
if _, err := fmt.Sscanf(s, "%d 0x%x", &x, &y); err != nil {
log.Fatalf("could not parse %q", s)
}
if x < 0 || 178*(0xc7-0x81)+(0xfe-0xc7)*94+(0xff-0xa1) <= x {
log.Fatalf("EUC-KR code %d is out of range", x)
}
mapping[x] = y
if reverse[y] == 0 {
c0, c1 := uint16(0), uint16(0)
if x < 178*(0xc7-0x81) {
c0 = uint16(x/178) + 0x81
c1 = uint16(x % 178)
switch {
case c1 < 1*26:
c1 += 0x41
case c1 < 2*26:
c1 += 0x47
default:
c1 += 0x4d
}
} else {
x -= 178 * (0xc7 - 0x81)
c0 = uint16(x/94) + 0xc7
c1 = uint16(x%94) + 0xa1
}
reverse[y] = c0<<8 | c1
}
}
if err := scanner.Err(); err != nil {
log.Fatalf("scanner error: %v", err)
}
fmt.Printf("// decode is the decoding table from EUC-KR code to Unicode.\n")
fmt.Printf("// It is defined at http://encoding.spec.whatwg.org/index-euc-kr.txt\n")
fmt.Printf("var decode = [...]uint16{\n")
for i, v := range mapping {
if v != 0 {
fmt.Printf("\t%d: 0x%04X,\n", i, v)
}
}
fmt.Printf("}\n\n")
// Any run of at least separation continuous zero entries in the reverse map will
// be a separate encode table.
const separation = 1024
intervals := []interval(nil)
low, high := -1, -1
for i, v := range reverse {
if v == 0 {
continue
}
if low < 0 {
low = i
} else if i-high >= separation {
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
low = i
}
high = i + 1
}
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
sort.Sort(byDecreasingLength(intervals))
fmt.Printf("const numEncodeTables = %d\n\n", len(intervals))
fmt.Printf("// encodeX are the encoding tables from Unicode to EUC-KR code,\n")
fmt.Printf("// sorted by decreasing length.\n")
for i, v := range intervals {
fmt.Printf("// encode%d: %5d entries for runes in [%5d, %5d).\n", i, v.len(), v.low, v.high)
}
fmt.Printf("\n")
for i, v := range intervals {
fmt.Printf("const encode%dLow, encode%dHigh = %d, %d\n\n", i, i, v.low, v.high)
fmt.Printf("var encode%d = [...]uint16{\n", i)
for j := v.low; j < v.high; j++ {
x := reverse[j]
if x == 0 {
continue
}
fmt.Printf("\t%d-%d: 0x%04X,\n", j, v.low, x)
}
fmt.Printf("}\n\n")
}
}
// interval is a half-open interval [low, high).
type interval struct {
low, high int
}
func (i interval) len() int { return i.high - i.low }
// byDecreasingLength sorts intervals by decreasing length.
type byDecreasingLength []interval
func (b byDecreasingLength) Len() int { return len(b) }
func (b byDecreasingLength) Less(i, j int) bool { return b[i].len() > b[j].len() }
func (b byDecreasingLength) Swap(i, j int) { b[i], b[j] = b[j], b[i] }

161
vendor/golang.org/x/text/encoding/simplifiedchinese/maketables.go generated vendored
View File

@ -1,161 +0,0 @@
// 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.
// +build ignore
package main
// This program generates tables.go:
// go run maketables.go | gofmt > tables.go
import (
"bufio"
"fmt"
"log"
"net/http"
"sort"
"strings"
)
func main() {
fmt.Printf("// generated by go run maketables.go; DO NOT EDIT\n\n")
fmt.Printf("// Package simplifiedchinese provides Simplified Chinese encodings such as GBK.\n")
fmt.Printf(`package simplifiedchinese // import "golang.org/x/text/encoding/simplifiedchinese"` + "\n\n")
printGB18030()
printGBK()
}
func printGB18030() {
res, err := http.Get("http://encoding.spec.whatwg.org/index-gb18030.txt")
if err != nil {
log.Fatalf("Get: %v", err)
}
defer res.Body.Close()
fmt.Printf("// gb18030 is the table from http://encoding.spec.whatwg.org/index-gb18030.txt\n")
fmt.Printf("var gb18030 = [...][2]uint16{\n")
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := uint32(0), uint32(0)
if _, err := fmt.Sscanf(s, "%d 0x%x", &x, &y); err != nil {
log.Fatalf("could not parse %q", s)
}
if x < 0x10000 && y < 0x10000 {
fmt.Printf("\t{0x%04x, 0x%04x},\n", x, y)
}
}
fmt.Printf("}\n\n")
}
func printGBK() {
res, err := http.Get("http://encoding.spec.whatwg.org/index-gbk.txt")
if err != nil {
log.Fatalf("Get: %v", err)
}
defer res.Body.Close()
mapping := [65536]uint16{}
reverse := [65536]uint16{}
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := uint16(0), uint16(0)
if _, err := fmt.Sscanf(s, "%d 0x%x", &x, &y); err != nil {
log.Fatalf("could not parse %q", s)
}
if x < 0 || 126*190 <= x {
log.Fatalf("GBK code %d is out of range", x)
}
mapping[x] = y
if reverse[y] == 0 {
c0, c1 := x/190, x%190
if c1 >= 0x3f {
c1++
}
reverse[y] = (0x81+c0)<<8 | (0x40 + c1)
}
}
if err := scanner.Err(); err != nil {
log.Fatalf("scanner error: %v", err)
}
fmt.Printf("// decode is the decoding table from GBK code to Unicode.\n")
fmt.Printf("// It is defined at http://encoding.spec.whatwg.org/index-gbk.txt\n")
fmt.Printf("var decode = [...]uint16{\n")
for i, v := range mapping {
if v != 0 {
fmt.Printf("\t%d: 0x%04X,\n", i, v)
}
}
fmt.Printf("}\n\n")
// Any run of at least separation continuous zero entries in the reverse map will
// be a separate encode table.
const separation = 1024
intervals := []interval(nil)
low, high := -1, -1
for i, v := range reverse {
if v == 0 {
continue
}
if low < 0 {
low = i
} else if i-high >= separation {
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
low = i
}
high = i + 1
}
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
sort.Sort(byDecreasingLength(intervals))
fmt.Printf("const numEncodeTables = %d\n\n", len(intervals))
fmt.Printf("// encodeX are the encoding tables from Unicode to GBK code,\n")
fmt.Printf("// sorted by decreasing length.\n")
for i, v := range intervals {
fmt.Printf("// encode%d: %5d entries for runes in [%5d, %5d).\n", i, v.len(), v.low, v.high)
}
fmt.Printf("\n")
for i, v := range intervals {
fmt.Printf("const encode%dLow, encode%dHigh = %d, %d\n\n", i, i, v.low, v.high)
fmt.Printf("var encode%d = [...]uint16{\n", i)
for j := v.low; j < v.high; j++ {
x := reverse[j]
if x == 0 {
continue
}
fmt.Printf("\t%d-%d: 0x%04X,\n", j, v.low, x)
}
fmt.Printf("}\n\n")
}
}
// interval is a half-open interval [low, high).
type interval struct {
low, high int
}
func (i interval) len() int { return i.high - i.low }
// byDecreasingLength sorts intervals by decreasing length.
type byDecreasingLength []interval
func (b byDecreasingLength) Len() int { return len(b) }
func (b byDecreasingLength) Less(i, j int) bool { return b[i].len() > b[j].len() }
func (b byDecreasingLength) Swap(i, j int) { b[i], b[j] = b[j], b[i] }

140
vendor/golang.org/x/text/encoding/traditionalchinese/maketables.go generated vendored
View File

@ -1,140 +0,0 @@
// 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.
// +build ignore
package main
// This program generates tables.go:
// go run maketables.go | gofmt > tables.go
import (
"bufio"
"fmt"
"log"
"net/http"
"sort"
"strings"
)
func main() {
fmt.Printf("// generated by go run maketables.go; DO NOT EDIT\n\n")
fmt.Printf("// Package traditionalchinese provides Traditional Chinese encodings such as Big5.\n")
fmt.Printf(`package traditionalchinese // import "golang.org/x/text/encoding/traditionalchinese"` + "\n\n")
res, err := http.Get("http://encoding.spec.whatwg.org/index-big5.txt")
if err != nil {
log.Fatalf("Get: %v", err)
}
defer res.Body.Close()
mapping := [65536]uint32{}
reverse := [65536 * 4]uint16{}
scanner := bufio.NewScanner(res.Body)
for scanner.Scan() {
s := strings.TrimSpace(scanner.Text())
if s == "" || s[0] == '#' {
continue
}
x, y := uint16(0), uint32(0)
if _, err := fmt.Sscanf(s, "%d 0x%x", &x, &y); err != nil {
log.Fatalf("could not parse %q", s)
}
if x < 0 || 126*157 <= x {
log.Fatalf("Big5 code %d is out of range", x)
}
mapping[x] = y
// The WHATWG spec http://encoding.spec.whatwg.org/#indexes says that
// "The index pointer for code point in index is the first pointer
// corresponding to code point in index", which would normally mean
// that the code below should be guarded by "if reverse[y] == 0", but
// last instead of first seems to match the behavior of
// "iconv -f UTF-8 -t BIG5". For example, U+8005 者 occurs twice in
// http://encoding.spec.whatwg.org/index-big5.txt, as index 2148
// (encoded as "\x8e\xcd") and index 6543 (encoded as "\xaa\xcc")
// and "echo 者 | iconv -f UTF-8 -t BIG5 | xxd" gives "\xaa\xcc".
c0, c1 := x/157, x%157
if c1 < 0x3f {
c1 += 0x40
} else {
c1 += 0x62
}
reverse[y] = (0x81+c0)<<8 | c1
}
if err := scanner.Err(); err != nil {
log.Fatalf("scanner error: %v", err)
}
fmt.Printf("// decode is the decoding table from Big5 code to Unicode.\n")
fmt.Printf("// It is defined at http://encoding.spec.whatwg.org/index-big5.txt\n")
fmt.Printf("var decode = [...]uint32{\n")
for i, v := range mapping {
if v != 0 {
fmt.Printf("\t%d: 0x%08X,\n", i, v)
}
}
fmt.Printf("}\n\n")
// Any run of at least separation continuous zero entries in the reverse map will
// be a separate encode table.
const separation = 1024
intervals := []interval(nil)
low, high := -1, -1
for i, v := range reverse {
if v == 0 {
continue
}
if low < 0 {
low = i
} else if i-high >= separation {
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
low = i
}
high = i + 1
}
if high >= 0 {
intervals = append(intervals, interval{low, high})
}
sort.Sort(byDecreasingLength(intervals))
fmt.Printf("const numEncodeTables = %d\n\n", len(intervals))
fmt.Printf("// encodeX are the encoding tables from Unicode to Big5 code,\n")
fmt.Printf("// sorted by decreasing length.\n")
for i, v := range intervals {
fmt.Printf("// encode%d: %5d entries for runes in [%6d, %6d).\n", i, v.len(), v.low, v.high)
}
fmt.Printf("\n")
for i, v := range intervals {
fmt.Printf("const encode%dLow, encode%dHigh = %d, %d\n\n", i, i, v.low, v.high)
fmt.Printf("var encode%d = [...]uint16{\n", i)
for j := v.low; j < v.high; j++ {
x := reverse[j]
if x == 0 {
continue
}
fmt.Printf("\t%d-%d: 0x%04X,\n", j, v.low, x)
}
fmt.Printf("}\n\n")
}
}
// interval is a half-open interval [low, high).
type interval struct {
low, high int
}
func (i interval) len() int { return i.high - i.low }
// byDecreasingLength sorts intervals by decreasing length.
type byDecreasingLength []interval
func (b byDecreasingLength) Len() int { return len(b) }
func (b byDecreasingLength) Less(i, j int) bool { return b[i].len() > b[j].len() }
func (b byDecreasingLength) Swap(i, j int) { b[i], b[j] = b[j], b[i] }

371
vendor/golang.org/x/text/internal/colltab/collelem.go generated vendored
View File

@ -1,371 +0,0 @@
// 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 colltab
import (
"fmt"
"unicode"
)
// Level identifies the collation comparison level.
// The primary level corresponds to the basic sorting of text.
// The secondary level corresponds to accents and related linguistic elements.
// The tertiary level corresponds to casing and related concepts.
// The quaternary level is derived from the other levels by the
// various algorithms for handling variable elements.
type Level int
const (
Primary Level = iota
Secondary
Tertiary
Quaternary
Identity
NumLevels
)
const (
defaultSecondary = 0x20
defaultTertiary = 0x2
maxTertiary = 0x1F
MaxQuaternary = 0x1FFFFF // 21 bits.
)
// Elem is a representation of a collation element. This API provides ways to encode
// and decode Elems. Implementations of collation tables may use values greater
// or equal to PrivateUse for their own purposes. However, these should never be
// returned by AppendNext.
type Elem uint32
const (
maxCE Elem = 0xAFFFFFFF
PrivateUse = minContract
minContract = 0xC0000000
maxContract = 0xDFFFFFFF
minExpand = 0xE0000000
maxExpand = 0xEFFFFFFF
minDecomp = 0xF0000000
)
type ceType int
const (
ceNormal ceType = iota // ceNormal includes implicits (ce == 0)
ceContractionIndex // rune can be a start of a contraction
ceExpansionIndex // rune expands into a sequence of collation elements
ceDecompose // rune expands using NFKC decomposition
)
func (ce Elem) ctype() ceType {
if ce <= maxCE {
return ceNormal
}
if ce <= maxContract {
return ceContractionIndex
} else {
if ce <= maxExpand {
return ceExpansionIndex
}
return ceDecompose
}
panic("should not reach here")
return ceType(-1)
}
// For normal collation elements, we assume that a collation element either has
// a primary or non-default secondary value, not both.
// Collation elements with a primary value are of the form
// 01pppppp pppppppp ppppppp0 ssssssss
// - p* is primary collation value
// - s* is the secondary collation value
// 00pppppp pppppppp ppppppps sssttttt, where
// - p* is primary collation value
// - s* offset of secondary from default value.
// - t* is the tertiary collation value
// 100ttttt cccccccc pppppppp pppppppp
// - t* is the tertiar collation value
// - c* is the canonical combining class
// - p* is the primary collation value
// Collation elements with a secondary value are of the form
// 1010cccc ccccssss ssssssss tttttttt, where
// - c* is the canonical combining class
// - s* is the secondary collation value
// - t* is the tertiary collation value
// 11qqqqqq qqqqqqqq qqqqqqq0 00000000
// - q* quaternary value
const (
ceTypeMask = 0xC0000000
ceTypeMaskExt = 0xE0000000
ceIgnoreMask = 0xF00FFFFF
ceType1 = 0x40000000
ceType2 = 0x00000000
ceType3or4 = 0x80000000
ceType4 = 0xA0000000
ceTypeQ = 0xC0000000
Ignore = ceType4
firstNonPrimary = 0x80000000
lastSpecialPrimary = 0xA0000000
secondaryMask = 0x80000000
hasTertiaryMask = 0x40000000
primaryValueMask = 0x3FFFFE00
maxPrimaryBits = 21
compactPrimaryBits = 16
maxSecondaryBits = 12
maxTertiaryBits = 8
maxCCCBits = 8
maxSecondaryCompactBits = 8
maxSecondaryDiffBits = 4
maxTertiaryCompactBits = 5
primaryShift = 9
compactSecondaryShift = 5
minCompactSecondary = defaultSecondary - 4
)
func makeImplicitCE(primary int) Elem {
return ceType1 | Elem(primary<<primaryShift) | defaultSecondary
}
// MakeElem returns an Elem for the given values. It will return an error
// if the given combination of values is invalid.
func MakeElem(primary, secondary, tertiary int, ccc uint8) (Elem, error) {
if w := primary; w >= 1<<maxPrimaryBits || w < 0 {
return 0, fmt.Errorf("makeCE: primary weight out of bounds: %x >= %x", w, 1<<maxPrimaryBits)
}
if w := secondary; w >= 1<<maxSecondaryBits || w < 0 {
return 0, fmt.Errorf("makeCE: secondary weight out of bounds: %x >= %x", w, 1<<maxSecondaryBits)
}
if w := tertiary; w >= 1<<maxTertiaryBits || w < 0 {
return 0, fmt.Errorf("makeCE: tertiary weight out of bounds: %x >= %x", w, 1<<maxTertiaryBits)
}
ce := Elem(0)
if primary != 0 {
if ccc != 0 {
if primary >= 1<<compactPrimaryBits {
return 0, fmt.Errorf("makeCE: primary weight with non-zero CCC out of bounds: %x >= %x", primary, 1<<compactPrimaryBits)
}
if secondary != defaultSecondary {
return 0, fmt.Errorf("makeCE: cannot combine non-default secondary value (%x) with non-zero CCC (%x)", secondary, ccc)
}
ce = Elem(tertiary << (compactPrimaryBits + maxCCCBits))
ce |= Elem(ccc) << compactPrimaryBits
ce |= Elem(primary)
ce |= ceType3or4
} else if tertiary == defaultTertiary {
if secondary >= 1<<maxSecondaryCompactBits {
return 0, fmt.Errorf("makeCE: secondary weight with non-zero primary out of bounds: %x >= %x", secondary, 1<<maxSecondaryCompactBits)
}
ce = Elem(primary<<(maxSecondaryCompactBits+1) + secondary)
ce |= ceType1
} else {
d := secondary - defaultSecondary + maxSecondaryDiffBits
if d >= 1<<maxSecondaryDiffBits || d < 0 {
return 0, fmt.Errorf("makeCE: secondary weight diff out of bounds: %x < 0 || %x > %x", d, d, 1<<maxSecondaryDiffBits)
}
if tertiary >= 1<<maxTertiaryCompactBits {
return 0, fmt.Errorf("makeCE: tertiary weight with non-zero primary out of bounds: %x > %x", tertiary, 1<<maxTertiaryCompactBits)
}
ce = Elem(primary<<maxSecondaryDiffBits + d)
ce = ce<<maxTertiaryCompactBits + Elem(tertiary)
}
} else {
ce = Elem(secondary<<maxTertiaryBits + tertiary)
ce += Elem(ccc) << (maxSecondaryBits + maxTertiaryBits)
ce |= ceType4
}
return ce, nil
}
// MakeQuaternary returns an Elem with the given quaternary value.
func MakeQuaternary(v int) Elem {
return ceTypeQ | Elem(v<<primaryShift)
}
// Mask sets weights for any level smaller than l to 0.
// The resulting Elem can be used to test for equality with
// other Elems to which the same mask has been applied.
func (ce Elem) Mask(l Level) uint32 {
return 0
}
// CCC returns the canonical combining class associated with the underlying character,
// if applicable, or 0 otherwise.
func (ce Elem) CCC() uint8 {
if ce&ceType3or4 != 0 {
if ce&ceType4 == ceType3or4 {
return uint8(ce >> 16)
}
return uint8(ce >> 20)
}
return 0
}
// Primary returns the primary collation weight for ce.
func (ce Elem) Primary() int {
if ce >= firstNonPrimary {
if ce > lastSpecialPrimary {
return 0
}
return int(uint16(ce))
}
return int(ce&primaryValueMask) >> primaryShift
}
// Secondary returns the secondary collation weight for ce.
func (ce Elem) Secondary() int {
switch ce & ceTypeMask {
case ceType1:
return int(uint8(ce))
case ceType2:
return minCompactSecondary + int((ce>>compactSecondaryShift)&0xF)
case ceType3or4:
if ce < ceType4 {
return defaultSecondary
}
return int(ce>>8) & 0xFFF
case ceTypeQ:
return 0
}
panic("should not reach here")
}
// Tertiary returns the tertiary collation weight for ce.
func (ce Elem) Tertiary() uint8 {
if ce&hasTertiaryMask == 0 {
if ce&ceType3or4 == 0 {
return uint8(ce & 0x1F)
}
if ce&ceType4 == ceType4 {
return uint8(ce)
}
return uint8(ce>>24) & 0x1F // type 2
} else if ce&ceTypeMask == ceType1 {
return defaultTertiary
}
// ce is a quaternary value.
return 0
}
func (ce Elem) updateTertiary(t uint8) Elem {
if ce&ceTypeMask == ceType1 {
// convert to type 4
nce := ce & primaryValueMask
nce |= Elem(uint8(ce)-minCompactSecondary) << compactSecondaryShift
ce = nce
} else if ce&ceTypeMaskExt == ceType3or4 {
ce &= ^Elem(maxTertiary << 24)
return ce | (Elem(t) << 24)
} else {
// type 2 or 4
ce &= ^Elem(maxTertiary)
}
return ce | Elem(t)
}
// Quaternary returns the quaternary value if explicitly specified,
// 0 if ce == Ignore, or MaxQuaternary otherwise.
// Quaternary values are used only for shifted variants.
func (ce Elem) Quaternary() int {
if ce&ceTypeMask == ceTypeQ {
return int(ce&primaryValueMask) >> primaryShift
} else if ce&ceIgnoreMask == Ignore {
return 0
}
return MaxQuaternary
}
// Weight returns the collation weight for the given level.
func (ce Elem) Weight(l Level) int {
switch l {
case Primary:
return ce.Primary()
case Secondary:
return ce.Secondary()
case Tertiary:
return int(ce.Tertiary())
case Quaternary:
return ce.Quaternary()
}
return 0 // return 0 (ignore) for undefined levels.
}
// For contractions, collation elements are of the form
// 110bbbbb bbbbbbbb iiiiiiii iiiinnnn, where
// - n* is the size of the first node in the contraction trie.
// - i* is the index of the first node in the contraction trie.
// - b* is the offset into the contraction collation element table.
// See contract.go for details on the contraction trie.
const (
maxNBits = 4
maxTrieIndexBits = 12
maxContractOffsetBits = 13
)
func splitContractIndex(ce Elem) (index, n, offset int) {
n = int(ce & (1<<maxNBits - 1))
ce >>= maxNBits
index = int(ce & (1<<maxTrieIndexBits - 1))
ce >>= maxTrieIndexBits
offset = int(ce & (1<<maxContractOffsetBits - 1))
return
}
// For expansions, Elems are of the form 11100000 00000000 bbbbbbbb bbbbbbbb,
// where b* is the index into the expansion sequence table.
const maxExpandIndexBits = 16
func splitExpandIndex(ce Elem) (index int) {
return int(uint16(ce))
}
// Some runes can be expanded using NFKD decomposition. Instead of storing the full
// sequence of collation elements, we decompose the rune and lookup the collation
// elements for each rune in the decomposition and modify the tertiary weights.
// The Elem, in this case, is of the form 11110000 00000000 wwwwwwww vvvvvvvv, where
// - v* is the replacement tertiary weight for the first rune,
// - w* is the replacement tertiary weight for the second rune,
// Tertiary weights of subsequent runes should be replaced with maxTertiary.
// See https://www.unicode.org/reports/tr10/#Compatibility_Decompositions for more details.
func splitDecompose(ce Elem) (t1, t2 uint8) {
return uint8(ce), uint8(ce >> 8)
}
const (
// These constants were taken from https://www.unicode.org/versions/Unicode6.0.0/ch12.pdf.
minUnified rune = 0x4E00
maxUnified = 0x9FFF
minCompatibility = 0xF900
maxCompatibility = 0xFAFF
minRare = 0x3400
maxRare = 0x4DBF
)
const (
commonUnifiedOffset = 0x10000
rareUnifiedOffset = 0x20000 // largest rune in common is U+FAFF
otherOffset = 0x50000 // largest rune in rare is U+2FA1D
illegalOffset = otherOffset + int(unicode.MaxRune)
maxPrimary = illegalOffset + 1
)
// implicitPrimary returns the primary weight for the a rune
// for which there is no entry for the rune in the collation table.
// We take a different approach from the one specified in
// https://unicode.org/reports/tr10/#Implicit_Weights,
// but preserve the resulting relative ordering of the runes.
func implicitPrimary(r rune) int {
if unicode.Is(unicode.Ideographic, r) {
if r >= minUnified && r <= maxUnified {
// The most common case for CJK.
return int(r) + commonUnifiedOffset
}
if r >= minCompatibility && r <= maxCompatibility {
// This will typically not hit. The DUCET explicitly specifies mappings
// for all characters that do not decompose.
return int(r) + commonUnifiedOffset
}
return int(r) + rareUnifiedOffset
}
return int(r) + otherOffset
}

105
vendor/golang.org/x/text/internal/colltab/colltab.go generated vendored
View File

@ -1,105 +0,0 @@
// Copyright 2015 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 colltab contains functionality related to collation tables.
// It is only to be used by the collate and search packages.
package colltab // import "golang.org/x/text/internal/colltab"
import (
"sort"
"golang.org/x/text/language"
)
// MatchLang finds the index of t in tags, using a matching algorithm used for
// collation and search. tags[0] must be language.Und, the remaining tags should
// be sorted alphabetically.
//
// Language matching for collation and search is different from the matching
// defined by language.Matcher: the (inferred) base language must be an exact
// match for the relevant fields. For example, "gsw" should not match "de".
// Also the parent relation is different, as a parent may have a different
// script. So usually the parent of zh-Hant is und, whereas for MatchLang it is
// zh.
func MatchLang(t language.Tag, tags []language.Tag) int {
// Canonicalize the values, including collapsing macro languages.
t, _ = language.All.Canonicalize(t)
base, conf := t.Base()
// Estimate the base language, but only use high-confidence values.
if conf < language.High {
// The root locale supports "search" and "standard". We assume that any
// implementation will only use one of both.
return 0
}
// Maximize base and script and normalize the tag.
if _, s, r := t.Raw(); (r != language.Region{}) {
p, _ := language.Raw.Compose(base, s, r)
// Taking the parent forces the script to be maximized.
p = p.Parent()
// Add back region and extensions.
t, _ = language.Raw.Compose(p, r, t.Extensions())
} else {
// Set the maximized base language.
t, _ = language.Raw.Compose(base, s, t.Extensions())
}
// Find start index of the language tag.
start := 1 + sort.Search(len(tags)-1, func(i int) bool {
b, _, _ := tags[i+1].Raw()
return base.String() <= b.String()
})
if start < len(tags) {
if b, _, _ := tags[start].Raw(); b != base {
return 0
}
}
// Besides the base language, script and region, only the collation type and
// the custom variant defined in the 'u' extension are used to distinguish a
// locale.
// Strip all variants and extensions and add back the custom variant.
tdef, _ := language.Raw.Compose(t.Raw())
tdef, _ = tdef.SetTypeForKey("va", t.TypeForKey("va"))
// First search for a specialized collation type, if present.
try := []language.Tag{tdef}
if co := t.TypeForKey("co"); co != "" {
tco, _ := tdef.SetTypeForKey("co", co)
try = []language.Tag{tco, tdef}
}
for _, tx := range try {
for ; tx != language.Und; tx = parent(tx) {
for i, t := range tags[start:] {
if b, _, _ := t.Raw(); b != base {
break
}
if tx == t {
return start + i
}
}
}
}
return 0
}
// parent computes the structural parent. This means inheritance may change
// script. So, unlike the CLDR parent, parent(zh-Hant) == zh.
func parent(t language.Tag) language.Tag {
if t.TypeForKey("va") != "" {
t, _ = t.SetTypeForKey("va", "")
return t
}
result := language.Und
if b, s, r := t.Raw(); (r != language.Region{}) {
result, _ = language.Raw.Compose(b, s, t.Extensions())
} else if (s != language.Script{}) {
result, _ = language.Raw.Compose(b, t.Extensions())
} else if (b != language.Base{}) {
result, _ = language.Raw.Compose(t.Extensions())
}
return result
}

145
vendor/golang.org/x/text/internal/colltab/contract.go generated vendored
View File

@ -1,145 +0,0 @@
// 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 colltab
import "unicode/utf8"
// For a description of ContractTrieSet, see text/collate/build/contract.go.
type ContractTrieSet []struct{ L, H, N, I uint8 }
// ctScanner is used to match a trie to an input sequence.
// A contraction may match a non-contiguous sequence of bytes in an input string.
// For example, if there is a contraction for <a, combining_ring>, it should match
// the sequence <a, combining_cedilla, combining_ring>, as combining_cedilla does
// not block combining_ring.
// ctScanner does not automatically skip over non-blocking non-starters, but rather
// retains the state of the last match and leaves it up to the user to continue
// the match at the appropriate points.
type ctScanner struct {
states ContractTrieSet
s []byte
n int
index int
pindex int
done bool
}
type ctScannerString struct {
states ContractTrieSet
s string
n int
index int
pindex int
done bool
}
func (t ContractTrieSet) scanner(index, n int, b []byte) ctScanner {
return ctScanner{s: b, states: t[index:], n: n}
}
func (t ContractTrieSet) scannerString(index, n int, str string) ctScannerString {
return ctScannerString{s: str, states: t[index:], n: n}
}
// result returns the offset i and bytes consumed p so far. If no suffix
// matched, i and p will be 0.
func (s *ctScanner) result() (i, p int) {
return s.index, s.pindex
}
func (s *ctScannerString) result() (i, p int) {
return s.index, s.pindex
}
const (
final = 0
noIndex = 0xFF
)
// scan matches the longest suffix at the current location in the input
// and returns the number of bytes consumed.
func (s *ctScanner) scan(p int) int {
pr := p // the p at the rune start
str := s.s
states, n := s.states, s.n
for i := 0; i < n && p < len(str); {
e := states[i]
c := str[p]
// TODO: a significant number of contractions are of a form that
// cannot match discontiguous UTF-8 in a normalized string. We could let
// a negative value of e.n mean that we can set s.done = true and avoid
// the need for additional matches.
if c >= e.L {
if e.L == c {
p++
if e.I != noIndex {
s.index = int(e.I)
s.pindex = p
}
if e.N != final {
i, states, n = 0, states[int(e.H)+n:], int(e.N)
if p >= len(str) || utf8.RuneStart(str[p]) {
s.states, s.n, pr = states, n, p
}
} else {
s.done = true
return p
}
continue
} else if e.N == final && c <= e.H {
p++
s.done = true
s.index = int(c-e.L) + int(e.I)
s.pindex = p
return p
}
}
i++
}
return pr
}
// scan is a verbatim copy of ctScanner.scan.
func (s *ctScannerString) scan(p int) int {
pr := p // the p at the rune start
str := s.s
states, n := s.states, s.n
for i := 0; i < n && p < len(str); {
e := states[i]
c := str[p]
// TODO: a significant number of contractions are of a form that
// cannot match discontiguous UTF-8 in a normalized string. We could let
// a negative value of e.n mean that we can set s.done = true and avoid
// the need for additional matches.
if c >= e.L {
if e.L == c {
p++
if e.I != noIndex {
s.index = int(e.I)
s.pindex = p
}
if e.N != final {
i, states, n = 0, states[int(e.H)+n:], int(e.N)
if p >= len(str) || utf8.RuneStart(str[p]) {
s.states, s.n, pr = states, n, p
}
} else {
s.done = true
return p
}
continue
} else if e.N == final && c <= e.H {
p++
s.done = true
s.index = int(c-e.L) + int(e.I)
s.pindex = p
return p
}
}
i++
}
return pr
}

178
vendor/golang.org/x/text/internal/colltab/iter.go generated vendored
View File

@ -1,178 +0,0 @@
// Copyright 2015 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 colltab
// An Iter incrementally converts chunks of the input text to collation
// elements, while ensuring that the collation elements are in normalized order
// (that is, they are in the order as if the input text were normalized first).
type Iter struct {
Weighter Weighter
Elems []Elem
// N is the number of elements in Elems that will not be reordered on
// subsequent iterations, N <= len(Elems).
N int
bytes []byte
str string
// Because the Elems buffer may contain collation elements that are needed
// for look-ahead, we need two positions in the text (bytes or str): one for
// the end position in the text for the current iteration and one for the
// start of the next call to appendNext.
pEnd int // end position in text corresponding to N.
pNext int // pEnd <= pNext.
}
// Reset sets the position in the current input text to p and discards any
// results obtained so far.
func (i *Iter) Reset(p int) {
i.Elems = i.Elems[:0]
i.N = 0
i.pEnd = p
i.pNext = p
}
// Len returns the length of the input text.
func (i *Iter) Len() int {
if i.bytes != nil {
return len(i.bytes)
}
return len(i.str)
}
// Discard removes the collation elements up to N.
func (i *Iter) Discard() {
// TODO: change this such that only modifiers following starters will have
// to be copied.
i.Elems = i.Elems[:copy(i.Elems, i.Elems[i.N:])]
i.N = 0
}
// End returns the end position of the input text for which Next has returned
// results.
func (i *Iter) End() int {
return i.pEnd
}
// SetInput resets i to input s.
func (i *Iter) SetInput(s []byte) {
i.bytes = s
i.str = ""
i.Reset(0)
}
// SetInputString resets i to input s.
func (i *Iter) SetInputString(s string) {
i.str = s
i.bytes = nil
i.Reset(0)
}
func (i *Iter) done() bool {
return i.pNext >= len(i.str) && i.pNext >= len(i.bytes)
}
func (i *Iter) appendNext() bool {
if i.done() {
return false
}
var sz int
if i.bytes == nil {
i.Elems, sz = i.Weighter.AppendNextString(i.Elems, i.str[i.pNext:])
} else {
i.Elems, sz = i.Weighter.AppendNext(i.Elems, i.bytes[i.pNext:])
}
if sz == 0 {
sz = 1
}
i.pNext += sz
return true
}
// Next appends Elems to the internal array. On each iteration, it will either
// add starters or modifiers. In the majority of cases, an Elem with a primary
// value > 0 will have a CCC of 0. The CCC values of collation elements are also
// used to detect if the input string was not normalized and to adjust the
// result accordingly.
func (i *Iter) Next() bool {
if i.N == len(i.Elems) && !i.appendNext() {
return false
}
// Check if the current segment starts with a starter.
prevCCC := i.Elems[len(i.Elems)-1].CCC()
if prevCCC == 0 {
i.N = len(i.Elems)
i.pEnd = i.pNext
return true
} else if i.Elems[i.N].CCC() == 0 {
// set i.N to only cover part of i.Elems for which prevCCC == 0 and
// use rest for the next call to next.
for i.N++; i.N < len(i.Elems) && i.Elems[i.N].CCC() == 0; i.N++ {
}
i.pEnd = i.pNext
return true
}
// The current (partial) segment starts with modifiers. We need to collect
// all successive modifiers to ensure that they are normalized.
for {
p := len(i.Elems)
i.pEnd = i.pNext
if !i.appendNext() {
break
}
if ccc := i.Elems[p].CCC(); ccc == 0 || len(i.Elems)-i.N > maxCombiningCharacters {
// Leave the starter for the next iteration. This ensures that we
// do not return sequences of collation elements that cross two
// segments.
//
// TODO: handle large number of combining characters by fully
// normalizing the input segment before iteration. This ensures
// results are consistent across the text repo.
i.N = p
return true
} else if ccc < prevCCC {
i.doNorm(p, ccc) // should be rare, never occurs for NFD and FCC.
} else {
prevCCC = ccc
}
}
done := len(i.Elems) != i.N
i.N = len(i.Elems)
return done
}
// nextNoNorm is the same as next, but does not "normalize" the collation
// elements.
func (i *Iter) nextNoNorm() bool {
// TODO: remove this function. Using this instead of next does not seem
// to improve performance in any significant way. We retain this until
// later for evaluation purposes.
if i.done() {
return false
}
i.appendNext()
i.N = len(i.Elems)
return true
}
const maxCombiningCharacters = 30
// doNorm reorders the collation elements in i.Elems.
// It assumes that blocks of collation elements added with appendNext
// either start and end with the same CCC or start with CCC == 0.
// This allows for a single insertion point for the entire block.
// The correctness of this assumption is verified in builder.go.
func (i *Iter) doNorm(p int, ccc uint8) {
n := len(i.Elems)
k := p
for p--; p > i.N && ccc < i.Elems[p-1].CCC(); p-- {
}
i.Elems = append(i.Elems, i.Elems[p:k]...)
copy(i.Elems[p:], i.Elems[k:])
i.Elems = i.Elems[:n]
}

236
vendor/golang.org/x/text/internal/colltab/numeric.go generated vendored
View File

@ -1,236 +0,0 @@
// Copyright 2014 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 colltab
import (
"unicode"
"unicode/utf8"
)
// NewNumericWeighter wraps w to replace individual digits to sort based on their
// numeric value.
//
// Weighter w must have a free primary weight after the primary weight for 9.
// If this is not the case, numeric value will sort at the same primary level
// as the first primary sorting after 9.
func NewNumericWeighter(w Weighter) Weighter {
getElem := func(s string) Elem {
elems, _ := w.AppendNextString(nil, s)
return elems[0]
}
nine := getElem("9")
// Numbers should order before zero, but the DUCET has no room for this.
// TODO: move before zero once we use fractional collation elements.
ns, _ := MakeElem(nine.Primary()+1, nine.Secondary(), int(nine.Tertiary()), 0)
return &numericWeighter{
Weighter: w,
// We assume that w sorts digits of different kinds in order of numeric
// value and that the tertiary weight order is preserved.
//
// TODO: evaluate whether it is worth basing the ranges on the Elem
// encoding itself once the move to fractional weights is complete.
zero: getElem("0"),
zeroSpecialLo: getElem("０"), // U+FF10 FULLWIDTH DIGIT ZERO
zeroSpecialHi: getElem("₀"), // U+2080 SUBSCRIPT ZERO
nine: nine,
nineSpecialHi: getElem("₉"), // U+2089 SUBSCRIPT NINE
numberStart: ns,
}
}
// A numericWeighter translates a stream of digits into a stream of weights
// representing the numeric value.
type numericWeighter struct {
Weighter
// The Elems below all demarcate boundaries of specific ranges. With the
// current element encoding digits are in two ranges: normal (default
// tertiary value) and special. For most languages, digits have collation
// elements in the normal range.
//
// Note: the range tests are very specific for the element encoding used by
// this implementation. The tests in collate_test.go are designed to fail
// if this code is not updated when an encoding has changed.
zero Elem // normal digit zero
zeroSpecialLo Elem // special digit zero, low tertiary value
zeroSpecialHi Elem // special digit zero, high tertiary value
nine Elem // normal digit nine
nineSpecialHi Elem // special digit nine
numberStart Elem
}
// AppendNext calls the namesake of the underlying weigher, but replaces single
// digits with weights representing their value.
func (nw *numericWeighter) AppendNext(buf []Elem, s []byte) (ce []Elem, n int) {
ce, n = nw.Weighter.AppendNext(buf, s)
nc := numberConverter{
elems: buf,
w: nw,
b: s,
}
isZero, ok := nc.checkNextDigit(ce)
if !ok {
return ce, n
}
// ce might have been grown already, so take it instead of buf.
nc.init(ce, len(buf), isZero)
for n < len(s) {
ce, sz := nw.Weighter.AppendNext(nc.elems, s[n:])
nc.b = s
n += sz
if !nc.update(ce) {
break
}
}
return nc.result(), n
}
// AppendNextString calls the namesake of the underlying weigher, but replaces
// single digits with weights representing their value.
func (nw *numericWeighter) AppendNextString(buf []Elem, s string) (ce []Elem, n int) {
ce, n = nw.Weighter.AppendNextString(buf, s)
nc := numberConverter{
elems: buf,
w: nw,
s: s,
}
isZero, ok := nc.checkNextDigit(ce)
if !ok {
return ce, n
}
nc.init(ce, len(buf), isZero)
for n < len(s) {
ce, sz := nw.Weighter.AppendNextString(nc.elems, s[n:])
nc.s = s
n += sz
if !nc.update(ce) {
break
}
}
return nc.result(), n
}
type numberConverter struct {
w *numericWeighter
elems []Elem
nDigits int
lenIndex int
s string // set if the input was of type string
b []byte // set if the input was of type []byte
}
// init completes initialization of a numberConverter and prepares it for adding
// more digits. elems is assumed to have a digit starting at oldLen.
func (nc *numberConverter) init(elems []Elem, oldLen int, isZero bool) {
// Insert a marker indicating the start of a number and a placeholder
// for the number of digits.
if isZero {
elems = append(elems[:oldLen], nc.w.numberStart, 0)
} else {
elems = append(elems, 0, 0)
copy(elems[oldLen+2:], elems[oldLen:])
elems[oldLen] = nc.w.numberStart
elems[oldLen+1] = 0
nc.nDigits = 1
}
nc.elems = elems
nc.lenIndex = oldLen + 1
}
// checkNextDigit reports whether bufNew adds a single digit relative to the old
// buffer. If it does, it also reports whether this digit is zero.
func (nc *numberConverter) checkNextDigit(bufNew []Elem) (isZero, ok bool) {
if len(nc.elems) >= len(bufNew) {
return false, false
}
e := bufNew[len(nc.elems)]
if e < nc.w.zeroSpecialLo || nc.w.nine < e {
// Not a number.
return false, false
}
if e < nc.w.zero {
if e > nc.w.nineSpecialHi {
// Not a number.
return false, false
}
if !nc.isDigit() {
return false, false
}
isZero = e <= nc.w.zeroSpecialHi
} else {
// This is the common case if we encounter a digit.
isZero = e == nc.w.zero
}
// Test the remaining added collation elements have a zero primary value.
if n := len(bufNew) - len(nc.elems); n > 1 {
for i := len(nc.elems) + 1; i < len(bufNew); i++ {
if bufNew[i].Primary() != 0 {
return false, false
}
}
// In some rare cases, collation elements will encode runes in
// unicode.No as a digit. For example Ethiopic digits (U+1369 - U+1371)
// are not in Nd. Also some digits that clearly belong in unicode.No,
// like U+0C78 TELUGU FRACTION DIGIT ZERO FOR ODD POWERS OF FOUR, have
// collation elements indistinguishable from normal digits.
// Unfortunately, this means we need to make this check for nearly all
// non-Latin digits.
//
// TODO: check the performance impact and find something better if it is
// an issue.
if !nc.isDigit() {
return false, false
}
}
return isZero, true
}
func (nc *numberConverter) isDigit() bool {
if nc.b != nil {
r, _ := utf8.DecodeRune(nc.b)
return unicode.In(r, unicode.Nd)
}
r, _ := utf8.DecodeRuneInString(nc.s)
return unicode.In(r, unicode.Nd)
}
// We currently support a maximum of about 2M digits (the number of primary
// values). Such numbers will compare correctly against small numbers, but their
// comparison against other large numbers is undefined.
//
// TODO: define a proper fallback, such as comparing large numbers textually or
// actually allowing numbers of unlimited length.
//
// TODO: cap this to a lower number (like 100) and maybe allow a larger number
// in an option?
const maxDigits = 1<<maxPrimaryBits - 1
func (nc *numberConverter) update(elems []Elem) bool {
isZero, ok := nc.checkNextDigit(elems)
if nc.nDigits == 0 && isZero {
return true
}
nc.elems = elems
if !ok {
return false
}
nc.nDigits++
return nc.nDigits < maxDigits
}
// result fills in the length element for the digit sequence and returns the
// completed collation elements.
func (nc *numberConverter) result() []Elem {
e, _ := MakeElem(nc.nDigits, defaultSecondary, defaultTertiary, 0)
nc.elems[nc.lenIndex] = e
return nc.elems
}

275
vendor/golang.org/x/text/internal/colltab/table.go generated vendored
View File

@ -1,275 +0,0 @@
// 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 colltab
import (
"unicode/utf8"
"golang.org/x/text/unicode/norm"
)
// Table holds all collation data for a given collation ordering.
type Table struct {
Index Trie // main trie
// expansion info
ExpandElem []uint32
// contraction info
ContractTries ContractTrieSet
ContractElem []uint32
MaxContractLen int
VariableTop uint32
}
func (t *Table) AppendNext(w []Elem, b []byte) (res []Elem, n int) {
return t.appendNext(w, source{bytes: b})
}
func (t *Table) AppendNextString(w []Elem, s string) (res []Elem, n int) {
return t.appendNext(w, source{str: s})
}
func (t *Table) Start(p int, b []byte) int {
// TODO: implement
panic("not implemented")
}
func (t *Table) StartString(p int, s string) int {
// TODO: implement
panic("not implemented")
}
func (t *Table) Domain() []string {
// TODO: implement
panic("not implemented")
}
func (t *Table) Top() uint32 {
return t.VariableTop
}
type source struct {
str string
bytes []byte
}
func (src *source) lookup(t *Table) (ce Elem, sz int) {
if src.bytes == nil {
return t.Index.lookupString(src.str)
}
return t.Index.lookup(src.bytes)
}
func (src *source) tail(sz int) {
if src.bytes == nil {
src.str = src.str[sz:]
} else {
src.bytes = src.bytes[sz:]
}
}
func (src *source) nfd(buf []byte, end int) []byte {
if src.bytes == nil {
return norm.NFD.AppendString(buf[:0], src.str[:end])
}
return norm.NFD.Append(buf[:0], src.bytes[:end]...)
}
func (src *source) rune() (r rune, sz int) {
if src.bytes == nil {
return utf8.DecodeRuneInString(src.str)
}
return utf8.DecodeRune(src.bytes)
}
func (src *source) properties(f norm.Form) norm.Properties {
if src.bytes == nil {
return f.PropertiesString(src.str)
}
return f.Properties(src.bytes)
}
// appendNext appends the weights corresponding to the next rune or
// contraction in s. If a contraction is matched to a discontinuous
// sequence of runes, the weights for the interstitial runes are
// appended as well. It returns a new slice that includes the appended
// weights and the number of bytes consumed from s.
func (t *Table) appendNext(w []Elem, src source) (res []Elem, n int) {
ce, sz := src.lookup(t)
tp := ce.ctype()
if tp == ceNormal {
if ce == 0 {
r, _ := src.rune()
const (
hangulSize = 3
firstHangul = 0xAC00
lastHangul = 0xD7A3
)
if r >= firstHangul && r <= lastHangul {
// TODO: performance can be considerably improved here.
n = sz
var buf [16]byte // Used for decomposing Hangul.
for b := src.nfd(buf[:0], hangulSize); len(b) > 0; b = b[sz:] {
ce, sz = t.Index.lookup(b)
w = append(w, ce)
}
return w, n
}
ce = makeImplicitCE(implicitPrimary(r))
}
w = append(w, ce)
} else if tp == ceExpansionIndex {
w = t.appendExpansion(w, ce)
} else if tp == ceContractionIndex {
n := 0
src.tail(sz)
if src.bytes == nil {
w, n = t.matchContractionString(w, ce, src.str)
} else {
w, n = t.matchContraction(w, ce, src.bytes)
}
sz += n
} else if tp == ceDecompose {
// Decompose using NFKD and replace tertiary weights.
t1, t2 := splitDecompose(ce)
i := len(w)
nfkd := src.properties(norm.NFKD).Decomposition()
for p := 0; len(nfkd) > 0; nfkd = nfkd[p:] {
w, p = t.appendNext(w, source{bytes: nfkd})
}
w[i] = w[i].updateTertiary(t1)
if i++; i < len(w) {
w[i] = w[i].updateTertiary(t2)
for i++; i < len(w); i++ {
w[i] = w[i].updateTertiary(maxTertiary)
}
}
}
return w, sz
}
func (t *Table) appendExpansion(w []Elem, ce Elem) []Elem {
i := splitExpandIndex(ce)
n := int(t.ExpandElem[i])
i++
for _, ce := range t.ExpandElem[i : i+n] {
w = append(w, Elem(ce))
}
return w
}
func (t *Table) matchContraction(w []Elem, ce Elem, suffix []byte) ([]Elem, int) {
index, n, offset := splitContractIndex(ce)
scan := t.ContractTries.scanner(index, n, suffix)
buf := [norm.MaxSegmentSize]byte{}
bufp := 0
p := scan.scan(0)
if !scan.done && p < len(suffix) && suffix[p] >= utf8.RuneSelf {
// By now we should have filtered most cases.
p0 := p
bufn := 0
rune := norm.NFD.Properties(suffix[p:])
p += rune.Size()
if rune.LeadCCC() != 0 {
prevCC := rune.TrailCCC()
// A gap may only occur in the last normalization segment.
// This also ensures that len(scan.s) < norm.MaxSegmentSize.
if end := norm.NFD.FirstBoundary(suffix[p:]); end != -1 {
scan.s = suffix[:p+end]
}
for p < len(suffix) && !scan.done && suffix[p] >= utf8.RuneSelf {
rune = norm.NFD.Properties(suffix[p:])
if ccc := rune.LeadCCC(); ccc == 0 || prevCC >= ccc {
break
}
prevCC = rune.TrailCCC()
if pp := scan.scan(p); pp != p {
// Copy the interstitial runes for later processing.
bufn += copy(buf[bufn:], suffix[p0:p])
if scan.pindex == pp {
bufp = bufn
}
p, p0 = pp, pp
} else {
p += rune.Size()
}
}
}
}
// Append weights for the matched contraction, which may be an expansion.
i, n := scan.result()
ce = Elem(t.ContractElem[i+offset])
if ce.ctype() == ceNormal {
w = append(w, ce)
} else {
w = t.appendExpansion(w, ce)
}
// Append weights for the runes in the segment not part of the contraction.
for b, p := buf[:bufp], 0; len(b) > 0; b = b[p:] {
w, p = t.appendNext(w, source{bytes: b})
}
return w, n
}
// TODO: unify the two implementations. This is best done after first simplifying
// the algorithm taking into account the inclusion of both NFC and NFD forms
// in the table.
func (t *Table) matchContractionString(w []Elem, ce Elem, suffix string) ([]Elem, int) {
index, n, offset := splitContractIndex(ce)
scan := t.ContractTries.scannerString(index, n, suffix)
buf := [norm.MaxSegmentSize]byte{}
bufp := 0
p := scan.scan(0)
if !scan.done && p < len(suffix) && suffix[p] >= utf8.RuneSelf {
// By now we should have filtered most cases.
p0 := p
bufn := 0
rune := norm.NFD.PropertiesString(suffix[p:])
p += rune.Size()
if rune.LeadCCC() != 0 {
prevCC := rune.TrailCCC()
// A gap may only occur in the last normalization segment.
// This also ensures that len(scan.s) < norm.MaxSegmentSize.
if end := norm.NFD.FirstBoundaryInString(suffix[p:]); end != -1 {
scan.s = suffix[:p+end]
}
for p < len(suffix) && !scan.done && suffix[p] >= utf8.RuneSelf {
rune = norm.NFD.PropertiesString(suffix[p:])
if ccc := rune.LeadCCC(); ccc == 0 || prevCC >= ccc {
break
}
prevCC = rune.TrailCCC()
if pp := scan.scan(p); pp != p {
// Copy the interstitial runes for later processing.
bufn += copy(buf[bufn:], suffix[p0:p])
if scan.pindex == pp {
bufp = bufn
}
p, p0 = pp, pp
} else {
p += rune.Size()
}
}
}
}
// Append weights for the matched contraction, which may be an expansion.
i, n := scan.result()
ce = Elem(t.ContractElem[i+offset])
if ce.ctype() == ceNormal {
w = append(w, ce)
} else {
w = t.appendExpansion(w, ce)
}
// Append weights for the runes in the segment not part of the contraction.
for b, p := buf[:bufp], 0; len(b) > 0; b = b[p:] {
w, p = t.appendNext(w, source{bytes: b})
}
return w, n
}

159
vendor/golang.org/x/text/internal/colltab/trie.go generated vendored
View File

@ -1,159 +0,0 @@
// 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.
// The trie in this file is used to associate the first full character in an
// UTF-8 string to a collation element. All but the last byte in a UTF-8 byte
// sequence are used to lookup offsets in the index table to be used for the
// next byte. The last byte is used to index into a table of collation elements.
// For a full description, see go.text/collate/build/trie.go.
package colltab
const blockSize = 64
type Trie struct {
Index0 []uint16 // index for first byte (0xC0-0xFF)
Values0 []uint32 // index for first byte (0x00-0x7F)
Index []uint16
Values []uint32
}
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
t6 = 0xFC // 1111 1100
te = 0xFE // 1111 1110
)
func (t *Trie) lookupValue(n uint16, b byte) Elem {
return Elem(t.Values[int(n)<<6+int(b)])
}
// lookup returns the trie value for the first UTF-8 encoding in s and
// the width in bytes of this encoding. The size will be 0 if s does not
// hold enough bytes to complete the encoding. len(s) must be greater than 0.
func (t *Trie) lookup(s []byte) (v Elem, sz int) {
c0 := s[0]
switch {
case c0 < tx:
return Elem(t.Values0[c0]), 1
case c0 < t2:
return 0, 1
case c0 < t3:
if len(s) < 2 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
return t.lookupValue(i, c1), 2
case c0 < t4:
if len(s) < 3 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
o := int(i)<<6 + int(c1)
i = t.Index[o]
c2 := s[2]
if c2 < tx || t2 <= c2 {
return 0, 2
}
return t.lookupValue(i, c2), 3
case c0 < t5:
if len(s) < 4 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
o := int(i)<<6 + int(c1)
i = t.Index[o]
c2 := s[2]
if c2 < tx || t2 <= c2 {
return 0, 2
}
o = int(i)<<6 + int(c2)
i = t.Index[o]
c3 := s[3]
if c3 < tx || t2 <= c3 {
return 0, 3
}
return t.lookupValue(i, c3), 4
}
// Illegal rune
return 0, 1
}
// The body of lookupString is a verbatim copy of that of lookup.
func (t *Trie) lookupString(s string) (v Elem, sz int) {
c0 := s[0]
switch {
case c0 < tx:
return Elem(t.Values0[c0]), 1
case c0 < t2:
return 0, 1
case c0 < t3:
if len(s) < 2 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
return t.lookupValue(i, c1), 2
case c0 < t4:
if len(s) < 3 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
o := int(i)<<6 + int(c1)
i = t.Index[o]
c2 := s[2]
if c2 < tx || t2 <= c2 {
return 0, 2
}
return t.lookupValue(i, c2), 3
case c0 < t5:
if len(s) < 4 {
return 0, 0
}
i := t.Index0[c0]
c1 := s[1]
if c1 < tx || t2 <= c1 {
return 0, 1
}
o := int(i)<<6 + int(c1)
i = t.Index[o]
c2 := s[2]
if c2 < tx || t2 <= c2 {
return 0, 2
}
o = int(i)<<6 + int(c2)
i = t.Index[o]
c3 := s[3]
if c3 < tx || t2 <= c3 {
return 0, 3
}
return t.lookupValue(i, c3), 4
}
// Illegal rune
return 0, 1
}

31
vendor/golang.org/x/text/internal/colltab/weighter.go generated vendored
View File

@ -1,31 +0,0 @@
// 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 colltab // import "golang.org/x/text/internal/colltab"
// A Weighter can be used as a source for Collator and Searcher.
type Weighter interface {
// Start finds the start of the segment that includes position p.
Start(p int, b []byte) int
// StartString finds the start of the segment that includes position p.
StartString(p int, s string) int
// AppendNext appends Elems to buf corresponding to the longest match
// of a single character or contraction from the start of s.
// It returns the new buf and the number of bytes consumed.
AppendNext(buf []Elem, s []byte) (ce []Elem, n int)
// AppendNextString appends Elems to buf corresponding to the longest match
// of a single character or contraction from the start of s.
// It returns the new buf and the number of bytes consumed.
AppendNextString(buf []Elem, s string) (ce []Elem, n int)
// Domain returns a slice of all single characters and contractions for which
// collation elements are defined in this table.
Domain() []string
// Top returns the highest variable primary value.
Top() uint32
}

375
vendor/golang.org/x/text/internal/gen/code.go generated vendored
View File

@ -1,375 +0,0 @@
// Copyright 2015 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 gen
import (
"bytes"
"encoding/gob"
"fmt"
"hash"
"hash/fnv"
"io"
"log"
"os"
"reflect"
"strings"
"unicode"
"unicode/utf8"
)
// This file contains utilities for generating code.
// TODO: other write methods like:
// - slices, maps, types, etc.
// CodeWriter is a utility for writing structured code. It computes the content
// hash and size of written content. It ensures there are newlines between
// written code blocks.
type CodeWriter struct {
buf bytes.Buffer
Size int
Hash hash.Hash32 // content hash
gob *gob.Encoder
// For comments we skip the usual one-line separator if they are followed by
// a code block.
skipSep bool
}
func (w *CodeWriter) Write(p []byte) (n int, err error) {
return w.buf.Write(p)
}
// NewCodeWriter returns a new CodeWriter.
func NewCodeWriter() *CodeWriter {
h := fnv.New32()
return &CodeWriter{Hash: h, gob: gob.NewEncoder(h)}
}
// WriteGoFile appends the buffer with the total size of all created structures
// and writes it as a Go file to the given file with the given package name.
func (w *CodeWriter) WriteGoFile(filename, pkg string) {
f, err := os.Create(filename)
if err != nil {
log.Fatalf("Could not create file %s: %v", filename, err)
}
defer f.Close()
if _, err = w.WriteGo(f, pkg, ""); err != nil {
log.Fatalf("Error writing file %s: %v", filename, err)
}
}
// WriteVersionedGoFile appends the buffer with the total size of all created
// structures and writes it as a Go file to the given file with the given
// package name and build tags for the current Unicode version,
func (w *CodeWriter) WriteVersionedGoFile(filename, pkg string) {
tags := buildTags()
if tags != "" {
pattern := fileToPattern(filename)
updateBuildTags(pattern)
filename = fmt.Sprintf(pattern, UnicodeVersion())
}
f, err := os.Create(filename)
if err != nil {
log.Fatalf("Could not create file %s: %v", filename, err)
}
defer f.Close()
if _, err = w.WriteGo(f, pkg, tags); err != nil {
log.Fatalf("Error writing file %s: %v", filename, err)
}
}
// WriteGo appends the buffer with the total size of all created structures and
// writes it as a Go file to the given writer with the given package name.
func (w *CodeWriter) WriteGo(out io.Writer, pkg, tags string) (n int, err error) {
sz := w.Size
if sz > 0 {
w.WriteComment("Total table size %d bytes (%dKiB); checksum: %X\n", sz, sz/1024, w.Hash.Sum32())
}
defer w.buf.Reset()
return WriteGo(out, pkg, tags, w.buf.Bytes())
}
func (w *CodeWriter) printf(f string, x ...interface{}) {
fmt.Fprintf(w, f, x...)
}
func (w *CodeWriter) insertSep() {
if w.skipSep {
w.skipSep = false
return
}
// Use at least two newlines to ensure a blank space between the previous
// block. WriteGoFile will remove extraneous newlines.
w.printf("\n\n")
}
// WriteComment writes a comment block. All line starts are prefixed with "//".
// Initial empty lines are gobbled. The indentation for the first line is
// stripped from consecutive lines.
func (w *CodeWriter) WriteComment(comment string, args ...interface{}) {
s := fmt.Sprintf(comment, args...)
s = strings.Trim(s, "\n")
// Use at least two newlines to ensure a blank space between the previous
// block. WriteGoFile will remove extraneous newlines.
w.printf("\n\n// ")
w.skipSep = true
// strip first indent level.
sep := "\n"
for ; len(s) > 0 && (s[0] == '\t' || s[0] == ' '); s = s[1:] {
sep += s[:1]
}
strings.NewReplacer(sep, "\n// ", "\n", "\n// ").WriteString(w, s)
w.printf("\n")
}
func (w *CodeWriter) writeSizeInfo(size int) {
w.printf("// Size: %d bytes\n", size)
}
// WriteConst writes a constant of the given name and value.
func (w *CodeWriter) WriteConst(name string, x interface{}) {
w.insertSep()
v := reflect.ValueOf(x)
switch v.Type().Kind() {
case reflect.String:
w.printf("const %s %s = ", name, typeName(x))
w.WriteString(v.String())
w.printf("\n")
default:
w.printf("const %s = %#v\n", name, x)
}
}
// WriteVar writes a variable of the given name and value.
func (w *CodeWriter) WriteVar(name string, x interface{}) {
w.insertSep()
v := reflect.ValueOf(x)
oldSize := w.Size
sz := int(v.Type().Size())
w.Size += sz
switch v.Type().Kind() {
case reflect.String:
w.printf("var %s %s = ", name, typeName(x))
w.WriteString(v.String())
case reflect.Struct:
w.gob.Encode(x)
fallthrough
case reflect.Slice, reflect.Array:
w.printf("var %s = ", name)
w.writeValue(v)
w.writeSizeInfo(w.Size - oldSize)
default:
w.printf("var %s %s = ", name, typeName(x))
w.gob.Encode(x)
w.writeValue(v)
w.writeSizeInfo(w.Size - oldSize)
}
w.printf("\n")
}
func (w *CodeWriter) writeValue(v reflect.Value) {
x := v.Interface()
switch v.Kind() {
case reflect.String:
w.WriteString(v.String())
case reflect.Array:
// Don't double count: callers of WriteArray count on the size being
// added, so we need to discount it here.
w.Size -= int(v.Type().Size())
w.writeSlice(x, true)
case reflect.Slice:
w.writeSlice(x, false)
case reflect.Struct:
w.printf("%s{\n", typeName(v.Interface()))
t := v.Type()
for i := 0; i < v.NumField(); i++ {
w.printf("%s: ", t.Field(i).Name)
w.writeValue(v.Field(i))
w.printf(",\n")
}
w.printf("}")
default:
w.printf("%#v", x)
}
}
// WriteString writes a string literal.
func (w *CodeWriter) WriteString(s string) {
io.WriteString(w.Hash, s) // content hash
w.Size += len(s)
const maxInline = 40
if len(s) <= maxInline {
w.printf("%q", s)
return
}
// We will render the string as a multi-line string.
const maxWidth = 80 - 4 - len(`"`) - len(`" +`)
// When starting on its own line, go fmt indents line 2+ an extra level.
n, max := maxWidth, maxWidth-4
// As per https://golang.org/issue/18078, the compiler has trouble
// compiling the concatenation of many strings, s0 + s1 + s2 + ... + sN,
// for large N. We insert redundant, explicit parentheses to work around
// that, lowering the N at any given step: (s0 + s1 + ... + s63) + (s64 +
// ... + s127) + etc + (etc + ... + sN).
explicitParens, extraComment := len(s) > 128*1024, ""
if explicitParens {
w.printf(`(`)
extraComment = "; the redundant, explicit parens are for https://golang.org/issue/18078"
}
// Print "" +\n, if a string does not start on its own line.
b := w.buf.Bytes()
if p := len(bytes.TrimRight(b, " \t")); p > 0 && b[p-1] != '\n' {
w.printf("\"\" + // Size: %d bytes%s\n", len(s), extraComment)
n, max = maxWidth, maxWidth
}
w.printf(`"`)
for sz, p, nLines := 0, 0, 0; p < len(s); {
var r rune
r, sz = utf8.DecodeRuneInString(s[p:])
out := s[p : p+sz]
chars := 1
if !unicode.IsPrint(r) || r == utf8.RuneError || r == '"' {
switch sz {
case 1:
out = fmt.Sprintf("\\x%02x", s[p])
case 2, 3:
out = fmt.Sprintf("\\u%04x", r)
case 4:
out = fmt.Sprintf("\\U%08x", r)
}
chars = len(out)
} else if r == '\\' {
out = "\\" + string(r)
chars = 2
}
if n -= chars; n < 0 {
nLines++
if explicitParens && nLines&63 == 63 {
w.printf("\") + (\"")
}
w.printf("\" +\n\"")
n = max - len(out)
}
w.printf("%s", out)
p += sz
}
w.printf(`"`)
if explicitParens {
w.printf(`)`)
}
}
// WriteSlice writes a slice value.
func (w *CodeWriter) WriteSlice(x interface{}) {
w.writeSlice(x, false)
}
// WriteArray writes an array value.
func (w *CodeWriter) WriteArray(x interface{}) {
w.writeSlice(x, true)
}
func (w *CodeWriter) writeSlice(x interface{}, isArray bool) {
v := reflect.ValueOf(x)
w.gob.Encode(v.Len())
w.Size += v.Len() * int(v.Type().Elem().Size())
name := typeName(x)
if isArray {
name = fmt.Sprintf("[%d]%s", v.Len(), name[strings.Index(name, "]")+1:])
}
if isArray {
w.printf("%s{\n", name)
} else {
w.printf("%s{ // %d elements\n", name, v.Len())
}
switch kind := v.Type().Elem().Kind(); kind {
case reflect.String:
for _, s := range x.([]string) {
w.WriteString(s)
w.printf(",\n")
}
case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64,
reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64:
// nLine and nBlock are the number of elements per line and block.
nLine, nBlock, format := 8, 64, "%d,"
switch kind {
case reflect.Uint8:
format = "%#02x,"
case reflect.Uint16:
format = "%#04x,"
case reflect.Uint32:
nLine, nBlock, format = 4, 32, "%#08x,"
case reflect.Uint, reflect.Uint64:
nLine, nBlock, format = 4, 32, "%#016x,"
case reflect.Int8:
nLine = 16
}
n := nLine
for i := 0; i < v.Len(); i++ {
if i%nBlock == 0 && v.Len() > nBlock {
w.printf("// Entry %X - %X\n", i, i+nBlock-1)
}
x := v.Index(i).Interface()
w.gob.Encode(x)
w.printf(format, x)
if n--; n == 0 {
n = nLine
w.printf("\n")
}
}
w.printf("\n")
case reflect.Struct:
zero := reflect.Zero(v.Type().Elem()).Interface()
for i := 0; i < v.Len(); i++ {
x := v.Index(i).Interface()
w.gob.EncodeValue(v)
if !reflect.DeepEqual(zero, x) {
line := fmt.Sprintf("%#v,\n", x)
line = line[strings.IndexByte(line, '{'):]
w.printf("%d: ", i)
w.printf(line)
}
}
case reflect.Array:
for i := 0; i < v.Len(); i++ {
w.printf("%d: %#v,\n", i, v.Index(i).Interface())
}
default:
panic("gen: slice elem type not supported")
}
w.printf("}")
}
// WriteType writes a definition of the type of the given value and returns the
// type name.
func (w *CodeWriter) WriteType(x interface{}) string {
t := reflect.TypeOf(x)
w.printf("type %s struct {\n", t.Name())
for i := 0; i < t.NumField(); i++ {
w.printf("\t%s %s\n", t.Field(i).Name, t.Field(i).Type)
}
w.printf("}\n")
return t.Name()
}
// typeName returns the name of the go type of x.
func typeName(x interface{}) string {
t := reflect.ValueOf(x).Type()
return strings.Replace(fmt.Sprint(t), "main.", "", 1)
}

347
vendor/golang.org/x/text/internal/gen/gen.go generated vendored
View File

@ -1,347 +0,0 @@
// Copyright 2015 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 gen contains common code for the various code generation tools in the
// text repository. Its usage ensures consistency between tools.
//
// This package defines command line flags that are common to most generation
// tools. The flags allow for specifying specific Unicode and CLDR versions
// in the public Unicode data repository (https://www.unicode.org/Public).
//
// A local Unicode data mirror can be set through the flag -local or the
// environment variable UNICODE_DIR. The former takes precedence. The local
// directory should follow the same structure as the public repository.
//
// IANA data can also optionally be mirrored by putting it in the iana directory
// rooted at the top of the local mirror. Beware, though, that IANA data is not
// versioned. So it is up to the developer to use the right version.
package gen // import "golang.org/x/text/internal/gen"
import (
"bytes"
"flag"
"fmt"
"go/build"
"go/format"
"io"
"io/ioutil"
"log"
"net/http"
"os"
"path"
"path/filepath"
"regexp"
"strings"
"sync"
"unicode"
"golang.org/x/text/unicode/cldr"
)
var (
url = flag.String("url",
"https://www.unicode.org/Public",
"URL of Unicode database directory")
iana = flag.String("iana",
"http://www.iana.org",
"URL of the IANA repository")
unicodeVersion = flag.String("unicode",
getEnv("UNICODE_VERSION", unicode.Version),
"unicode version to use")
cldrVersion = flag.String("cldr",
getEnv("CLDR_VERSION", cldr.Version),
"cldr version to use")
)
func getEnv(name, def string) string {
if v := os.Getenv(name); v != "" {
return v
}
return def
}
// Init performs common initialization for a gen command. It parses the flags
// and sets up the standard logging parameters.
func Init() {
log.SetPrefix("")
log.SetFlags(log.Lshortfile)
flag.Parse()
}
const header = `// Code generated by running "go generate" in golang.org/x/text. DO NOT EDIT.
`
// UnicodeVersion reports the requested Unicode version.
func UnicodeVersion() string {
return *unicodeVersion
}
// CLDRVersion reports the requested CLDR version.
func CLDRVersion() string {
return *cldrVersion
}
var tags = []struct{ version, buildTags string }{
{"9.0.0", "!go1.10"},
{"10.0.0", "go1.10,!go1.13"},
{"11.0.0", "go1.13"},
}
// buildTags reports the build tags used for the current Unicode version.
func buildTags() string {
v := UnicodeVersion()
for _, e := range tags {
if e.version == v {
return e.buildTags
}
}
log.Fatalf("Unknown build tags for Unicode version %q.", v)
return ""
}
// IsLocal reports whether data files are available locally.
func IsLocal() bool {
dir, err := localReadmeFile()
if err != nil {
return false
}
if _, err = os.Stat(dir); err != nil {
return false
}
return true
}
// OpenUCDFile opens the requested UCD file. The file is specified relative to
// the public Unicode root directory. It will call log.Fatal if there are any
// errors.
func OpenUCDFile(file string) io.ReadCloser {
return openUnicode(path.Join(*unicodeVersion, "ucd", file))
}
// OpenCLDRCoreZip opens the CLDR core zip file. It will call log.Fatal if there
// are any errors.
func OpenCLDRCoreZip() io.ReadCloser {
return OpenUnicodeFile("cldr", *cldrVersion, "core.zip")
}
// OpenUnicodeFile opens the requested file of the requested category from the
// root of the Unicode data archive. The file is specified relative to the
// public Unicode root directory. If version is "", it will use the default
// Unicode version. It will call log.Fatal if there are any errors.
func OpenUnicodeFile(category, version, file string) io.ReadCloser {
if version == "" {
version = UnicodeVersion()
}
return openUnicode(path.Join(category, version, file))
}
// OpenIANAFile opens the requested IANA file. The file is specified relative
// to the IANA root, which is typically either http://www.iana.org or the
// iana directory in the local mirror. It will call log.Fatal if there are any
// errors.
func OpenIANAFile(path string) io.ReadCloser {
return Open(*iana, "iana", path)
}
var (
dirMutex sync.Mutex
localDir string
)
const permissions = 0755
func localReadmeFile() (string, error) {
p, err := build.Import("golang.org/x/text", "", build.FindOnly)
if err != nil {
return "", fmt.Errorf("Could not locate package: %v", err)
}
return filepath.Join(p.Dir, "DATA", "README"), nil
}
func getLocalDir() string {
dirMutex.Lock()
defer dirMutex.Unlock()
readme, err := localReadmeFile()
if err != nil {
log.Fatal(err)
}
dir := filepath.Dir(readme)
if _, err := os.Stat(readme); err != nil {
if err := os.MkdirAll(dir, permissions); err != nil {
log.Fatalf("Could not create directory: %v", err)
}
ioutil.WriteFile(readme, []byte(readmeTxt), permissions)
}
return dir
}
const readmeTxt = `Generated by golang.org/x/text/internal/gen. DO NOT EDIT.
This directory contains downloaded files used to generate the various tables
in the golang.org/x/text subrepo.
Note that the language subtag repo (iana/assignments/language-subtag-registry)
and all other times in the iana subdirectory are not versioned and will need
to be periodically manually updated. The easiest way to do this is to remove
the entire iana directory. This is mostly of concern when updating the language
package.
`
// Open opens subdir/path if a local directory is specified and the file exists,
// where subdir is a directory relative to the local root, or fetches it from
// urlRoot/path otherwise. It will call log.Fatal if there are any errors.
func Open(urlRoot, subdir, path string) io.ReadCloser {
file := filepath.Join(getLocalDir(), subdir, filepath.FromSlash(path))
return open(file, urlRoot, path)
}
func openUnicode(path string) io.ReadCloser {
file := filepath.Join(getLocalDir(), filepath.FromSlash(path))
return open(file, *url, path)
}
// TODO: automatically periodically update non-versioned files.
func open(file, urlRoot, path string) io.ReadCloser {
if f, err := os.Open(file); err == nil {
return f
}
r := get(urlRoot, path)
defer r.Close()
b, err := ioutil.ReadAll(r)
if err != nil {
log.Fatalf("Could not download file: %v", err)
}
os.MkdirAll(filepath.Dir(file), permissions)
if err := ioutil.WriteFile(file, b, permissions); err != nil {
log.Fatalf("Could not create file: %v", err)
}
return ioutil.NopCloser(bytes.NewReader(b))
}
func get(root, path string) io.ReadCloser {
url := root + "/" + path
fmt.Printf("Fetching %s...", url)
defer fmt.Println(" done.")
resp, err := http.Get(url)
if err != nil {
log.Fatalf("HTTP GET: %v", err)
}
if resp.StatusCode != 200 {
log.Fatalf("Bad GET status for %q: %q", url, resp.Status)
}
return resp.Body
}
// TODO: use Write*Version in all applicable packages.
// WriteUnicodeVersion writes a constant for the Unicode version from which the
// tables are generated.
func WriteUnicodeVersion(w io.Writer) {
fmt.Fprintf(w, "// UnicodeVersion is the Unicode version from which the tables in this package are derived.\n")
fmt.Fprintf(w, "const UnicodeVersion = %q\n\n", UnicodeVersion())
}
// WriteCLDRVersion writes a constant for the CLDR version from which the
// tables are generated.
func WriteCLDRVersion(w io.Writer) {
fmt.Fprintf(w, "// CLDRVersion is the CLDR version from which the tables in this package are derived.\n")
fmt.Fprintf(w, "const CLDRVersion = %q\n\n", CLDRVersion())
}
// WriteGoFile prepends a standard file comment and package statement to the
// given bytes, applies gofmt, and writes them to a file with the given name.
// It will call log.Fatal if there are any errors.
func WriteGoFile(filename, pkg string, b []byte) {
w, err := os.Create(filename)
if err != nil {
log.Fatalf("Could not create file %s: %v", filename, err)
}
defer w.Close()
if _, err = WriteGo(w, pkg, "", b); err != nil {
log.Fatalf("Error writing file %s: %v", filename, err)
}
}
func fileToPattern(filename string) string {
suffix := ".go"
if strings.HasSuffix(filename, "_test.go") {
suffix = "_test.go"
}
prefix := filename[:len(filename)-len(suffix)]
return fmt.Sprint(prefix, "%s", suffix)
}
func updateBuildTags(pattern string) {
for _, t := range tags {
oldFile := fmt.Sprintf(pattern, t.version)
b, err := ioutil.ReadFile(oldFile)
if err != nil {
continue
}
build := fmt.Sprintf("// +build %s", t.buildTags)
b = regexp.MustCompile(`// \+build .*`).ReplaceAll(b, []byte(build))
err = ioutil.WriteFile(oldFile, b, 0644)
if err != nil {
log.Fatal(err)
}
}
}
// WriteVersionedGoFile prepends a standard file comment, adds build tags to
// version the file for the current Unicode version, and package statement to
// the given bytes, applies gofmt, and writes them to a file with the given
// name. It will call log.Fatal if there are any errors.
func WriteVersionedGoFile(filename, pkg string, b []byte) {
pattern := fileToPattern(filename)
updateBuildTags(pattern)
filename = fmt.Sprintf(pattern, UnicodeVersion())
w, err := os.Create(filename)
if err != nil {
log.Fatalf("Could not create file %s: %v", filename, err)
}
defer w.Close()
if _, err = WriteGo(w, pkg, buildTags(), b); err != nil {
log.Fatalf("Error writing file %s: %v", filename, err)
}
}
// WriteGo prepends a standard file comment and package statement to the given
// bytes, applies gofmt, and writes them to w.
func WriteGo(w io.Writer, pkg, tags string, b []byte) (n int, err error) {
src := []byte(header)
if tags != "" {
src = append(src, fmt.Sprintf("// +build %s\n\n", tags)...)
}
src = append(src, fmt.Sprintf("package %s\n\n", pkg)...)
src = append(src, b...)
formatted, err := format.Source(src)
if err != nil {
// Print the generated code even in case of an error so that the
// returned error can be meaningfully interpreted.
n, _ = w.Write(src)
return n, err
}
return w.Write(formatted)
}
// Repackage rewrites a Go file from belonging to package main to belonging to
// the given package.
func Repackage(inFile, outFile, pkg string) {
src, err := ioutil.ReadFile(inFile)
if err != nil {
log.Fatalf("reading %s: %v", inFile, err)
}
const toDelete = "package main\n\n"
i := bytes.Index(src, []byte(toDelete))
if i < 0 {
log.Fatalf("Could not find %q in %s.", toDelete, inFile)
}
w := &bytes.Buffer{}
w.Write(src[i+len(toDelete):])
WriteGoFile(outFile, pkg, w.Bytes())
}

64
vendor/golang.org/x/text/internal/language/compact/gen.go generated vendored
View File

@ -1,64 +0,0 @@
// 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.
// +build ignore
// Language tag table generator.
// Data read from the web.
package main
import (
"flag"
"fmt"
"log"
"golang.org/x/text/internal/gen"
"golang.org/x/text/unicode/cldr"
)
var (
test = flag.Bool("test",
false,
"test existing tables; can be used to compare web data with package data.")
outputFile = flag.String("output",
"tables.go",
"output file for generated tables")
)
func main() {
gen.Init()
w := gen.NewCodeWriter()
defer w.WriteGoFile("tables.go", "compact")
fmt.Fprintln(w, `import "golang.org/x/text/internal/language"`)
b := newBuilder(w)
gen.WriteCLDRVersion(w)
b.writeCompactIndex()
}
type builder struct {
w *gen.CodeWriter
data *cldr.CLDR
supp *cldr.SupplementalData
}
func newBuilder(w *gen.CodeWriter) *builder {
r := gen.OpenCLDRCoreZip()
defer r.Close()
d := &cldr.Decoder{}
data, err := d.DecodeZip(r)
if err != nil {
log.Fatal(err)
}
b := builder{
w: w,
data: data,
supp: data.Supplemental(),
}
return &b
}

113
vendor/golang.org/x/text/internal/language/compact/gen_index.go generated vendored
View File

@ -1,113 +0,0 @@
// Copyright 2015 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
package main
// This file generates derivative tables based on the language package itself.
import (
"fmt"
"log"
"sort"
"strings"
"golang.org/x/text/internal/language"
)
// Compact indices:
// Note -va-X variants only apply to localization variants.
// BCP variants only ever apply to language.
// The only ambiguity between tags is with regions.
func (b *builder) writeCompactIndex() {
// Collect all language tags for which we have any data in CLDR.
m := map[language.Tag]bool{}
for _, lang := range b.data.Locales() {
// We include all locales unconditionally to be consistent with en_US.
// We want en_US, even though it has no data associated with it.
// TODO: put any of the languages for which no data exists at the end
// of the index. This allows all components based on ICU to use that
// as the cutoff point.
// if x := data.RawLDML(lang); false ||
// x.LocaleDisplayNames != nil ||
// x.Characters != nil ||
// x.Delimiters != nil ||
// x.Measurement != nil ||
// x.Dates != nil ||
// x.Numbers != nil ||
// x.Units != nil ||
// x.ListPatterns != nil ||
// x.Collations != nil ||
// x.Segmentations != nil ||
// x.Rbnf != nil ||
// x.Annotations != nil ||
// x.Metadata != nil {
// TODO: support POSIX natively, albeit non-standard.
tag := language.Make(strings.Replace(lang, "_POSIX", "-u-va-posix", 1))
m[tag] = true
// }
}
// TODO: plural rules are also defined for the deprecated tags:
// iw mo sh tl
// Consider removing these as compact tags.
// Include locales for plural rules, which uses a different structure.
for _, plurals := range b.supp.Plurals {
for _, rules := range plurals.PluralRules {
for _, lang := range strings.Split(rules.Locales, " ") {
m[language.Make(lang)] = true
}
}
}
var coreTags []language.CompactCoreInfo
var special []string
for t := range m {
if x := t.Extensions(); len(x) != 0 && fmt.Sprint(x) != "[u-va-posix]" {
log.Fatalf("Unexpected extension %v in %v", x, t)
}
if len(t.Variants()) == 0 && len(t.Extensions()) == 0 {
cci, ok := language.GetCompactCore(t)
if !ok {
log.Fatalf("Locale for non-basic language %q", t)
}
coreTags = append(coreTags, cci)
} else {
special = append(special, t.String())
}
}
w := b.w
sort.Slice(coreTags, func(i, j int) bool { return coreTags[i] < coreTags[j] })
sort.Strings(special)
w.WriteComment(`
NumCompactTags is the number of common tags. The maximum tag is
NumCompactTags-1.`)
w.WriteConst("NumCompactTags", len(m))
fmt.Fprintln(w, "const (")
for i, t := range coreTags {
fmt.Fprintf(w, "%s ID = %d\n", ident(t.Tag().String()), i)
}
for i, t := range special {
fmt.Fprintf(w, "%s ID = %d\n", ident(t), i+len(coreTags))
}
fmt.Fprintln(w, ")")
w.WriteVar("coreTags", coreTags)
w.WriteConst("specialTagsStr", strings.Join(special, " "))
}
func ident(s string) string {
return strings.Replace(s, "-", "", -1) + "Index"
}

54
vendor/golang.org/x/text/internal/language/compact/gen_parents.go generated vendored
View File

@ -1,54 +0,0 @@
// Copyright 2018 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
package main
import (
"log"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/language"
"golang.org/x/text/internal/language/compact"
"golang.org/x/text/unicode/cldr"
)
func main() {
r := gen.OpenCLDRCoreZip()
defer r.Close()
d := &cldr.Decoder{}
data, err := d.DecodeZip(r)
if err != nil {
log.Fatalf("DecodeZip: %v", err)
}
w := gen.NewCodeWriter()
defer w.WriteGoFile("parents.go", "compact")
// Create parents table.
type ID uint16
parents := make([]ID, compact.NumCompactTags)
for _, loc := range data.Locales() {
tag := language.MustParse(loc)
index, ok := compact.FromTag(tag)
if !ok {
continue
}
parentIndex := compact.ID(0) // und
for p := tag.Parent(); p != language.Und; p = p.Parent() {
if x, ok := compact.FromTag(p); ok {
parentIndex = x
break
}
}
parents[index] = ID(parentIndex)
}
w.WriteComment(`
parents maps a compact index of a tag to the compact index of the parent of
this tag.`)
w.WriteVar("parents", parents)
}

1520
vendor/golang.org/x/text/internal/language/gen.go generated vendored
View File

File diff suppressed because it is too large Load Diff

20
vendor/golang.org/x/text/internal/language/gen_common.go generated vendored
View File

@ -1,20 +0,0 @@
// Copyright 2014 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
package main
// This file contains code common to the maketables.go and the package code.
// AliasType is the type of an alias in AliasMap.
type AliasType int8
const (
Deprecated AliasType = iota
Macro
Legacy
AliasTypeUnknown AliasType = -1
)

58
vendor/golang.org/x/text/internal/triegen/compact.go generated vendored
View File

@ -1,58 +0,0 @@
// Copyright 2014 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 triegen
// This file defines Compacter and its implementations.
import "io"
// A Compacter generates an alternative, more space-efficient way to store a
// trie value block. A trie value block holds all possible values for the last
// byte of a UTF-8 encoded rune. Excluding ASCII characters, a trie value block
// always has 64 values, as a UTF-8 encoding ends with a byte in [0x80, 0xC0).
type Compacter interface {
// Size returns whether the Compacter could encode the given block as well
// as its size in case it can. len(v) is always 64.
Size(v []uint64) (sz int, ok bool)
// Store stores the block using the Compacter's compression method.
// It returns a handle with which the block can be retrieved.
// len(v) is always 64.
Store(v []uint64) uint32
// Print writes the data structures associated to the given store to w.
Print(w io.Writer) error
// Handler returns the name of a function that gets called during trie
// lookup for blocks generated by the Compacter. The function should be of
// the form func (n uint32, b byte) uint64, where n is the index returned by
// the Compacter's Store method and b is the last byte of the UTF-8
// encoding, where 0x80 <= b < 0xC0, for which to do the lookup in the
// block.
Handler() string
}
// simpleCompacter is the default Compacter used by builder. It implements a
// normal trie block.
type simpleCompacter builder
func (b *simpleCompacter) Size([]uint64) (sz int, ok bool) {
return blockSize * b.ValueSize, true
}
func (b *simpleCompacter) Store(v []uint64) uint32 {
h := uint32(len(b.ValueBlocks) - blockOffset)
b.ValueBlocks = append(b.ValueBlocks, v)
return h
}
func (b *simpleCompacter) Print(io.Writer) error {
// Structures are printed in print.go.
return nil
}
func (b *simpleCompacter) Handler() string {
panic("Handler should be special-cased for this Compacter")
}

251
vendor/golang.org/x/text/internal/triegen/print.go generated vendored
View File

@ -1,251 +0,0 @@
// Copyright 2014 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 triegen
import (
"bytes"
"fmt"
"io"
"strings"
"text/template"
)
// print writes all the data structures as well as the code necessary to use the
// trie to w.
func (b *builder) print(w io.Writer) error {
b.Stats.NValueEntries = len(b.ValueBlocks) * blockSize
b.Stats.NValueBytes = len(b.ValueBlocks) * blockSize * b.ValueSize
b.Stats.NIndexEntries = len(b.IndexBlocks) * blockSize
b.Stats.NIndexBytes = len(b.IndexBlocks) * blockSize * b.IndexSize
b.Stats.NHandleBytes = len(b.Trie) * 2 * b.IndexSize
// If we only have one root trie, all starter blocks are at position 0 and
// we can access the arrays directly.
if len(b.Trie) == 1 {
// At this point we cannot refer to the generated tables directly.
b.ASCIIBlock = b.Name + "Values"
b.StarterBlock = b.Name + "Index"
} else {
// Otherwise we need to have explicit starter indexes in the trie
// structure.
b.ASCIIBlock = "t.ascii"
b.StarterBlock = "t.utf8Start"
}
b.SourceType = "[]byte"
if err := lookupGen.Execute(w, b); err != nil {
return err
}
b.SourceType = "string"
if err := lookupGen.Execute(w, b); err != nil {
return err
}
if err := trieGen.Execute(w, b); err != nil {
return err
}
for _, c := range b.Compactions {
if err := c.c.Print(w); err != nil {
return err
}
}
return nil
}
func printValues(n int, values []uint64) string {
w := &bytes.Buffer{}
boff := n * blockSize
fmt.Fprintf(w, "\t// Block %#x, offset %#x", n, boff)
var newline bool
for i, v := range values {
if i%6 == 0 {
newline = true
}
if v != 0 {
if newline {
fmt.Fprintf(w, "\n")
newline = false
}
fmt.Fprintf(w, "\t%#02x:%#04x, ", boff+i, v)
}
}
return w.String()
}
func printIndex(b *builder, nr int, n *node) string {
w := &bytes.Buffer{}
boff := nr * blockSize
fmt.Fprintf(w, "\t// Block %#x, offset %#x", nr, boff)
var newline bool
for i, c := range n.children {
if i%8 == 0 {
newline = true
}
if c != nil {
v := b.Compactions[c.index.compaction].Offset + uint32(c.index.index)
if v != 0 {
if newline {
fmt.Fprintf(w, "\n")
newline = false
}
fmt.Fprintf(w, "\t%#02x:%#02x, ", boff+i, v)
}
}
}
return w.String()
}
var (
trieGen = template.Must(template.New("trie").Funcs(template.FuncMap{
"printValues": printValues,
"printIndex": printIndex,
"title": strings.Title,
"dec": func(x int) int { return x - 1 },
"psize": func(n int) string {
return fmt.Sprintf("%d bytes (%.2f KiB)", n, float64(n)/1024)
},
}).Parse(trieTemplate))
lookupGen = template.Must(template.New("lookup").Parse(lookupTemplate))
)
// TODO: consider the return type of lookup. It could be uint64, even if the
// internal value type is smaller. We will have to verify this with the
// performance of unicode/norm, which is very sensitive to such changes.
const trieTemplate = `{{$b := .}}{{$multi := gt (len .Trie) 1}}
// {{.Name}}Trie. Total size: {{psize .Size}}. Checksum: {{printf "%08x" .Checksum}}.
type {{.Name}}Trie struct { {{if $multi}}
ascii []{{.ValueType}} // index for ASCII bytes
utf8Start []{{.IndexType}} // index for UTF-8 bytes >= 0xC0
{{end}}}
func new{{title .Name}}Trie(i int) *{{.Name}}Trie { {{if $multi}}
h := {{.Name}}TrieHandles[i]
return &{{.Name}}Trie{ {{.Name}}Values[uint32(h.ascii)<<6:], {{.Name}}Index[uint32(h.multi)<<6:] }
}
type {{.Name}}TrieHandle struct {
ascii, multi {{.IndexType}}
}
// {{.Name}}TrieHandles: {{len .Trie}} handles, {{.Stats.NHandleBytes}} bytes
var {{.Name}}TrieHandles = [{{len .Trie}}]{{.Name}}TrieHandle{
{{range .Trie}} { {{.ASCIIIndex}}, {{.StarterIndex}} }, // {{printf "%08x" .Checksum}}: {{.Name}}
{{end}}}{{else}}
return &{{.Name}}Trie{}
}
{{end}}
// lookupValue determines the type of block n and looks up the value for b.
func (t *{{.Name}}Trie) lookupValue(n uint32, b byte) {{.ValueType}}{{$last := dec (len .Compactions)}} {
switch { {{range $i, $c := .Compactions}}
{{if eq $i $last}}default{{else}}case n < {{$c.Cutoff}}{{end}}:{{if ne $i 0}}
n -= {{$c.Offset}}{{end}}
return {{print $b.ValueType}}({{$c.Handler}}){{end}}
}
}
// {{.Name}}Values: {{len .ValueBlocks}} blocks, {{.Stats.NValueEntries}} entries, {{.Stats.NValueBytes}} bytes
// The third block is the zero block.
var {{.Name}}Values = [{{.Stats.NValueEntries}}]{{.ValueType}} {
{{range $i, $v := .ValueBlocks}}{{printValues $i $v}}
{{end}}}
// {{.Name}}Index: {{len .IndexBlocks}} blocks, {{.Stats.NIndexEntries}} entries, {{.Stats.NIndexBytes}} bytes
// Block 0 is the zero block.
var {{.Name}}Index = [{{.Stats.NIndexEntries}}]{{.IndexType}} {
{{range $i, $v := .IndexBlocks}}{{printIndex $b $i $v}}
{{end}}}
`
// TODO: consider allowing zero-length strings after evaluating performance with
// unicode/norm.
const lookupTemplate = `
// lookup{{if eq .SourceType "string"}}String{{end}} returns the trie value for the first UTF-8 encoding in s and
// the width in bytes of this encoding. The size will be 0 if s does not
// hold enough bytes to complete the encoding. len(s) must be greater than 0.
func (t *{{.Name}}Trie) lookup{{if eq .SourceType "string"}}String{{end}}(s {{.SourceType}}) (v {{.ValueType}}, sz int) {
c0 := s[0]
switch {
case c0 < 0x80: // is ASCII
return {{.ASCIIBlock}}[c0], 1
case c0 < 0xC2:
return 0, 1 // Illegal UTF-8: not a starter, not ASCII.
case c0 < 0xE0: // 2-byte UTF-8
if len(s) < 2 {
return 0, 0
}
i := {{.StarterBlock}}[c0]
c1 := s[1]
if c1 < 0x80 || 0xC0 <= c1 {
return 0, 1 // Illegal UTF-8: not a continuation byte.
}
return t.lookupValue(uint32(i), c1), 2
case c0 < 0xF0: // 3-byte UTF-8
if len(s) < 3 {
return 0, 0
}
i := {{.StarterBlock}}[c0]
c1 := s[1]
if c1 < 0x80 || 0xC0 <= c1 {
return 0, 1 // Illegal UTF-8: not a continuation byte.
}
o := uint32(i)<<6 + uint32(c1)
i = {{.Name}}Index[o]
c2 := s[2]
if c2 < 0x80 || 0xC0 <= c2 {
return 0, 2 // Illegal UTF-8: not a continuation byte.
}
return t.lookupValue(uint32(i), c2), 3
case c0 < 0xF8: // 4-byte UTF-8
if len(s) < 4 {
return 0, 0
}
i := {{.StarterBlock}}[c0]
c1 := s[1]
if c1 < 0x80 || 0xC0 <= c1 {
return 0, 1 // Illegal UTF-8: not a continuation byte.
}
o := uint32(i)<<6 + uint32(c1)
i = {{.Name}}Index[o]
c2 := s[2]
if c2 < 0x80 || 0xC0 <= c2 {
return 0, 2 // Illegal UTF-8: not a continuation byte.
}
o = uint32(i)<<6 + uint32(c2)
i = {{.Name}}Index[o]
c3 := s[3]
if c3 < 0x80 || 0xC0 <= c3 {
return 0, 3 // Illegal UTF-8: not a continuation byte.
}
return t.lookupValue(uint32(i), c3), 4
}
// Illegal rune
return 0, 1
}
// lookup{{if eq .SourceType "string"}}String{{end}}Unsafe returns the trie value for the first UTF-8 encoding in s.
// s must start with a full and valid UTF-8 encoded rune.
func (t *{{.Name}}Trie) lookup{{if eq .SourceType "string"}}String{{end}}Unsafe(s {{.SourceType}}) {{.ValueType}} {
c0 := s[0]
if c0 < 0x80 { // is ASCII
return {{.ASCIIBlock}}[c0]
}
i := {{.StarterBlock}}[c0]
if c0 < 0xE0 { // 2-byte UTF-8
return t.lookupValue(uint32(i), s[1])
}
i = {{.Name}}Index[uint32(i)<<6+uint32(s[1])]
if c0 < 0xF0 { // 3-byte UTF-8
return t.lookupValue(uint32(i), s[2])
}
i = {{.Name}}Index[uint32(i)<<6+uint32(s[2])]
if c0 < 0xF8 { // 4-byte UTF-8
return t.lookupValue(uint32(i), s[3])
}
return 0
}
`

494
vendor/golang.org/x/text/internal/triegen/triegen.go generated vendored
View File

@ -1,494 +0,0 @@
// Copyright 2014 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 triegen implements a code generator for a trie for associating
// unsigned integer values with UTF-8 encoded runes.
//
// Many of the go.text packages use tries for storing per-rune information. A
// trie is especially useful if many of the runes have the same value. If this
// is the case, many blocks can be expected to be shared allowing for
// information on many runes to be stored in little space.
//
// As most of the lookups are done directly on []byte slices, the tries use the
// UTF-8 bytes directly for the lookup. This saves a conversion from UTF-8 to
// runes and contributes a little bit to better performance. It also naturally
// provides a fast path for ASCII.
//
// Space is also an issue. There are many code points defined in Unicode and as
// a result tables can get quite large. So every byte counts. The triegen
// package automatically chooses the smallest integer values to represent the
// tables. Compacters allow further compression of the trie by allowing for
// alternative representations of individual trie blocks.
//
// triegen allows generating multiple tries as a single structure. This is
// useful when, for example, one wants to generate tries for several languages
// that have a lot of values in common. Some existing libraries for
// internationalization store all per-language data as a dynamically loadable
// chunk. The go.text packages are designed with the assumption that the user
// typically wants to compile in support for all supported languages, in line
// with the approach common to Go to create a single standalone binary. The
// multi-root trie approach can give significant storage savings in this
// scenario.
//
// triegen generates both tables and code. The code is optimized to use the
// automatically chosen data types. The following code is generated for a Trie
// or multiple Tries named "foo":
// - type fooTrie
// The trie type.
//
// - func newFooTrie(x int) *fooTrie
// Trie constructor, where x is the index of the trie passed to Gen.
//
// - func (t *fooTrie) lookup(s []byte) (v uintX, sz int)
// The lookup method, where uintX is automatically chosen.
//
// - func lookupString, lookupUnsafe and lookupStringUnsafe
// Variants of the above.
//
// - var fooValues and fooIndex and any tables generated by Compacters.
// The core trie data.
//
// - var fooTrieHandles
// Indexes of starter blocks in case of multiple trie roots.
//
// It is recommended that users test the generated trie by checking the returned
// value for every rune. Such exhaustive tests are possible as the number of
// runes in Unicode is limited.
package triegen // import "golang.org/x/text/internal/triegen"
// TODO: Arguably, the internally optimized data types would not have to be
// exposed in the generated API. We could also investigate not generating the
// code, but using it through a package. We would have to investigate the impact
// on performance of making such change, though. For packages like unicode/norm,
// small changes like this could tank performance.
import (
"encoding/binary"
"fmt"
"hash/crc64"
"io"
"log"
"unicode/utf8"
)
// builder builds a set of tries for associating values with runes. The set of
// tries can share common index and value blocks.
type builder struct {
Name string
// ValueType is the type of the trie values looked up.
ValueType string
// ValueSize is the byte size of the ValueType.
ValueSize int
// IndexType is the type of trie index values used for all UTF-8 bytes of
// a rune except the last one.
IndexType string
// IndexSize is the byte size of the IndexType.
IndexSize int
// SourceType is used when generating the lookup functions. If the user
// requests StringSupport, all lookup functions will be generated for
// string input as well.
SourceType string
Trie []*Trie
IndexBlocks []*node
ValueBlocks [][]uint64
Compactions []compaction
Checksum uint64
ASCIIBlock string
StarterBlock string
indexBlockIdx map[uint64]int
valueBlockIdx map[uint64]nodeIndex
asciiBlockIdx map[uint64]int
// Stats are used to fill out the template.
Stats struct {
NValueEntries int
NValueBytes int
NIndexEntries int
NIndexBytes int
NHandleBytes int
}
err error
}
// A nodeIndex encodes the index of a node, which is defined by the compaction
// which stores it and an index within the compaction. For internal nodes, the
// compaction is always 0.
type nodeIndex struct {
compaction int
index int
}
// compaction keeps track of stats used for the compaction.
type compaction struct {
c Compacter
blocks []*node
maxHandle uint32
totalSize int
// Used by template-based generator and thus exported.
Cutoff uint32
Offset uint32
Handler string
}
func (b *builder) setError(err error) {
if b.err == nil {
b.err = err
}
}
// An Option can be passed to Gen.
type Option func(b *builder) error
// Compact configures the trie generator to use the given Compacter.
func Compact(c Compacter) Option {
return func(b *builder) error {
b.Compactions = append(b.Compactions, compaction{
c: c,
Handler: c.Handler() + "(n, b)"})
return nil
}
}
// Gen writes Go code for a shared trie lookup structure to w for the given
// Tries. The generated trie type will be called nameTrie. newNameTrie(x) will
// return the *nameTrie for tries[x]. A value can be looked up by using one of
// the various lookup methods defined on nameTrie. It returns the table size of
// the generated trie.
func Gen(w io.Writer, name string, tries []*Trie, opts ...Option) (sz int, err error) {
// The index contains two dummy blocks, followed by the zero block. The zero
// block is at offset 0x80, so that the offset for the zero block for
// continuation bytes is 0.
b := &builder{
Name: name,
Trie: tries,
IndexBlocks: []*node{{}, {}, {}},
Compactions: []compaction{{
Handler: name + "Values[n<<6+uint32(b)]",
}},
// The 0 key in indexBlockIdx and valueBlockIdx is the hash of the zero
// block.
indexBlockIdx: map[uint64]int{0: 0},
valueBlockIdx: map[uint64]nodeIndex{0: {}},
asciiBlockIdx: map[uint64]int{},
}
b.Compactions[0].c = (*simpleCompacter)(b)
for _, f := range opts {
if err := f(b); err != nil {
return 0, err
}
}
b.build()
if b.err != nil {
return 0, b.err
}
if err = b.print(w); err != nil {
return 0, err
}
return b.Size(), nil
}
// A Trie represents a single root node of a trie. A builder may build several
// overlapping tries at once.
type Trie struct {
root *node
hiddenTrie
}
// hiddenTrie contains values we want to be visible to the template generator,
// but hidden from the API documentation.
type hiddenTrie struct {
Name string
Checksum uint64
ASCIIIndex int
StarterIndex int
}
// NewTrie returns a new trie root.
func NewTrie(name string) *Trie {
return &Trie{
&node{
children: make([]*node, blockSize),
values: make([]uint64, utf8.RuneSelf),
},
hiddenTrie{Name: name},
}
}
// Gen is a convenience wrapper around the Gen func passing t as the only trie
// and uses the name passed to NewTrie. It returns the size of the generated
// tables.
func (t *Trie) Gen(w io.Writer, opts ...Option) (sz int, err error) {
return Gen(w, t.Name, []*Trie{t}, opts...)
}
// node is a node of the intermediate trie structure.
type node struct {
// children holds this node's children. It is always of length 64.
// A child node may be nil.
children []*node
// values contains the values of this node. If it is non-nil, this node is
// either a root or leaf node:
// For root nodes, len(values) == 128 and it maps the bytes in [0x00, 0x7F].
// For leaf nodes, len(values) == 64 and it maps the bytes in [0x80, 0xBF].
values []uint64
index nodeIndex
}
// Insert associates value with the given rune. Insert will panic if a non-zero
// value is passed for an invalid rune.
func (t *Trie) Insert(r rune, value uint64) {
if value == 0 {
return
}
s := string(r)
if []rune(s)[0] != r && value != 0 {
// Note: The UCD tables will always assign what amounts to a zero value
// to a surrogate. Allowing a zero value for an illegal rune allows
// users to iterate over [0..MaxRune] without having to explicitly
// exclude surrogates, which would be tedious.
panic(fmt.Sprintf("triegen: non-zero value for invalid rune %U", r))
}
if len(s) == 1 {
// It is a root node value (ASCII).
t.root.values[s[0]] = value
return
}
n := t.root
for ; len(s) > 1; s = s[1:] {
if n.children == nil {
n.children = make([]*node, blockSize)
}
p := s[0] % blockSize
c := n.children[p]
if c == nil {
c = &node{}
n.children[p] = c
}
if len(s) > 2 && c.values != nil {
log.Fatalf("triegen: insert(%U): found internal node with values", r)
}
n = c
}
if n.values == nil {
n.values = make([]uint64, blockSize)
}
if n.children != nil {
log.Fatalf("triegen: insert(%U): found leaf node that also has child nodes", r)
}
n.values[s[0]-0x80] = value
}
// Size returns the number of bytes the generated trie will take to store. It
// needs to be exported as it is used in the templates.
func (b *builder) Size() int {
// Index blocks.
sz := len(b.IndexBlocks) * blockSize * b.IndexSize
// Skip the first compaction, which represents the normal value blocks, as
// its totalSize does not account for the ASCII blocks, which are managed
// separately.
sz += len(b.ValueBlocks) * blockSize * b.ValueSize
for _, c := range b.Compactions[1:] {
sz += c.totalSize
}
// TODO: this computation does not account for the fixed overhead of a using
// a compaction, either code or data. As for data, though, the typical
// overhead of data is in the order of bytes (2 bytes for cases). Further,
// the savings of using a compaction should anyway be substantial for it to
// be worth it.
// For multi-root tries, we also need to account for the handles.
if len(b.Trie) > 1 {
sz += 2 * b.IndexSize * len(b.Trie)
}
return sz
}
func (b *builder) build() {
// Compute the sizes of the values.
var vmax uint64
for _, t := range b.Trie {
vmax = maxValue(t.root, vmax)
}
b.ValueType, b.ValueSize = getIntType(vmax)
// Compute all block allocations.
// TODO: first compute the ASCII blocks for all tries and then the other
// nodes. ASCII blocks are more restricted in placement, as they require two
// blocks to be placed consecutively. Processing them first may improve
// sharing (at least one zero block can be expected to be saved.)
for _, t := range b.Trie {
b.Checksum += b.buildTrie(t)
}
// Compute the offsets for all the Compacters.
offset := uint32(0)
for i := range b.Compactions {
c := &b.Compactions[i]
c.Offset = offset
offset += c.maxHandle + 1
c.Cutoff = offset
}
// Compute the sizes of indexes.
// TODO: different byte positions could have different sizes. So far we have
// not found a case where this is beneficial.
imax := uint64(b.Compactions[len(b.Compactions)-1].Cutoff)
for _, ib := range b.IndexBlocks {
if x := uint64(ib.index.index); x > imax {
imax = x
}
}
b.IndexType, b.IndexSize = getIntType(imax)
}
func maxValue(n *node, max uint64) uint64 {
if n == nil {
return max
}
for _, c := range n.children {
max = maxValue(c, max)
}
for _, v := range n.values {
if max < v {
max = v
}
}
return max
}
func getIntType(v uint64) (string, int) {
switch {
case v < 1<<8:
return "uint8", 1
case v < 1<<16:
return "uint16", 2
case v < 1<<32:
return "uint32", 4
}
return "uint64", 8
}
const (
blockSize = 64
// Subtract two blocks to offset 0x80, the first continuation byte.
blockOffset = 2
// Subtract three blocks to offset 0xC0, the first non-ASCII starter.
rootBlockOffset = 3
)
var crcTable = crc64.MakeTable(crc64.ISO)
func (b *builder) buildTrie(t *Trie) uint64 {
n := t.root
// Get the ASCII offset. For the first trie, the ASCII block will be at
// position 0.
hasher := crc64.New(crcTable)
binary.Write(hasher, binary.BigEndian, n.values)
hash := hasher.Sum64()
v, ok := b.asciiBlockIdx[hash]
if !ok {
v = len(b.ValueBlocks)
b.asciiBlockIdx[hash] = v
b.ValueBlocks = append(b.ValueBlocks, n.values[:blockSize], n.values[blockSize:])
if v == 0 {
// Add the zero block at position 2 so that it will be assigned a
// zero reference in the lookup blocks.
// TODO: always do this? This would allow us to remove a check from
// the trie lookup, but at the expense of extra space. Analyze
// performance for unicode/norm.
b.ValueBlocks = append(b.ValueBlocks, make([]uint64, blockSize))
}
}
t.ASCIIIndex = v
// Compute remaining offsets.
t.Checksum = b.computeOffsets(n, true)
// We already subtracted the normal blockOffset from the index. Subtract the
// difference for starter bytes.
t.StarterIndex = n.index.index - (rootBlockOffset - blockOffset)
return t.Checksum
}
func (b *builder) computeOffsets(n *node, root bool) uint64 {
// For the first trie, the root lookup block will be at position 3, which is
// the offset for UTF-8 non-ASCII starter bytes.
first := len(b.IndexBlocks) == rootBlockOffset
if first {
b.IndexBlocks = append(b.IndexBlocks, n)
}
// We special-case the cases where all values recursively are 0. This allows
// for the use of a zero block to which all such values can be directed.
hash := uint64(0)
if n.children != nil || n.values != nil {
hasher := crc64.New(crcTable)
for _, c := range n.children {
var v uint64
if c != nil {
v = b.computeOffsets(c, false)
}
binary.Write(hasher, binary.BigEndian, v)
}
binary.Write(hasher, binary.BigEndian, n.values)
hash = hasher.Sum64()
}
if first {
b.indexBlockIdx[hash] = rootBlockOffset - blockOffset
}
// Compacters don't apply to internal nodes.
if n.children != nil {
v, ok := b.indexBlockIdx[hash]
if !ok {
v = len(b.IndexBlocks) - blockOffset
b.IndexBlocks = append(b.IndexBlocks, n)
b.indexBlockIdx[hash] = v
}
n.index = nodeIndex{0, v}
} else {
h, ok := b.valueBlockIdx[hash]
if !ok {
bestI, bestSize := 0, blockSize*b.ValueSize
for i, c := range b.Compactions[1:] {
if sz, ok := c.c.Size(n.values); ok && bestSize > sz {
bestI, bestSize = i+1, sz
}
}
c := &b.Compactions[bestI]
c.totalSize += bestSize
v := c.c.Store(n.values)
if c.maxHandle < v {
c.maxHandle = v
}
h = nodeIndex{bestI, int(v)}
b.valueBlockIdx[hash] = h
}
n.index = h
}
return hash
}

371
vendor/golang.org/x/text/internal/ucd/ucd.go generated vendored
View File

@ -1,371 +0,0 @@
// Copyright 2014 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 ucd provides a parser for Unicode Character Database files, the
// format of which is defined in https://www.unicode.org/reports/tr44/. See
// https://www.unicode.org/Public/UCD/latest/ucd/ for example files.
//
// It currently does not support substitutions of missing fields.
package ucd // import "golang.org/x/text/internal/ucd"
import (
"bufio"
"errors"
"fmt"
"io"
"log"
"regexp"
"strconv"
"strings"
)
// UnicodeData.txt fields.
const (
CodePoint = iota
Name
GeneralCategory
CanonicalCombiningClass
BidiClass
DecompMapping
DecimalValue
DigitValue
NumericValue
BidiMirrored
Unicode1Name
ISOComment
SimpleUppercaseMapping
SimpleLowercaseMapping
SimpleTitlecaseMapping
)
// Parse calls f for each entry in the given reader of a UCD file. It will close
// the reader upon return. It will call log.Fatal if any error occurred.
//
// This implements the most common usage pattern of using Parser.
func Parse(r io.ReadCloser, f func(p *Parser)) {
defer r.Close()
p := New(r)
for p.Next() {
f(p)
}
if err := p.Err(); err != nil {
r.Close() // os.Exit will cause defers not to be called.
log.Fatal(err)
}
}
// An Option is used to configure a Parser.
type Option func(p *Parser)
func keepRanges(p *Parser) {
p.keepRanges = true
}
var (
// KeepRanges prevents the expansion of ranges. The raw ranges can be
// obtained by calling Range(0) on the parser.
KeepRanges Option = keepRanges
)
// The Part option register a handler for lines starting with a '@'. The text
// after a '@' is available as the first field. Comments are handled as usual.
func Part(f func(p *Parser)) Option {
return func(p *Parser) {
p.partHandler = f
}
}
// The CommentHandler option passes comments that are on a line by itself to
// a given handler.
func CommentHandler(f func(s string)) Option {
return func(p *Parser) {
p.commentHandler = f
}
}
// A Parser parses Unicode Character Database (UCD) files.
type Parser struct {
scanner *bufio.Scanner
keepRanges bool // Don't expand rune ranges in field 0.
err error
comment string
field []string
// parsedRange is needed in case Range(0) is called more than once for one
// field. In some cases this requires scanning ahead.
line int
parsedRange bool
rangeStart, rangeEnd rune
partHandler func(p *Parser)
commentHandler func(s string)
}
func (p *Parser) setError(err error, msg string) {
if p.err == nil && err != nil {
if msg == "" {
p.err = fmt.Errorf("ucd:line:%d: %v", p.line, err)
} else {
p.err = fmt.Errorf("ucd:line:%d:%s: %v", p.line, msg, err)
}
}
}
func (p *Parser) getField(i int) string {
if i >= len(p.field) {
return ""
}
return p.field[i]
}
// Err returns a non-nil error if any error occurred during parsing.
func (p *Parser) Err() error {
return p.err
}
// New returns a Parser for the given Reader.
func New(r io.Reader, o ...Option) *Parser {
p := &Parser{
scanner: bufio.NewScanner(r),
}
for _, f := range o {
f(p)
}
return p
}
// Next parses the next line in the file. It returns true if a line was parsed
// and false if it reached the end of the file.
func (p *Parser) Next() bool {
if !p.keepRanges && p.rangeStart < p.rangeEnd {
p.rangeStart++
return true
}
p.comment = ""
p.field = p.field[:0]
p.parsedRange = false
for p.scanner.Scan() && p.err == nil {
p.line++
s := p.scanner.Text()
if s == "" {
continue
}
if s[0] == '#' {
if p.commentHandler != nil {
p.commentHandler(strings.TrimSpace(s[1:]))
}
continue
}
// Parse line
if i := strings.IndexByte(s, '#'); i != -1 {
p.comment = strings.TrimSpace(s[i+1:])
s = s[:i]
}
if s[0] == '@' {
if p.partHandler != nil {
p.field = append(p.field, strings.TrimSpace(s[1:]))
p.partHandler(p)
p.field = p.field[:0]
}
p.comment = ""
continue
}
for {
i := strings.IndexByte(s, ';')
if i == -1 {
p.field = append(p.field, strings.TrimSpace(s))
break
}
p.field = append(p.field, strings.TrimSpace(s[:i]))
s = s[i+1:]
}
if !p.keepRanges {
p.rangeStart, p.rangeEnd = p.getRange(0)
}
return true
}
p.setError(p.scanner.Err(), "scanner failed")
return false
}
func parseRune(b string) (rune, error) {
if len(b) > 2 && b[0] == 'U' && b[1] == '+' {
b = b[2:]
}
x, err := strconv.ParseUint(b, 16, 32)
return rune(x), err
}
func (p *Parser) parseRune(s string) rune {
x, err := parseRune(s)
p.setError(err, "failed to parse rune")
return x
}
// Rune parses and returns field i as a rune.
func (p *Parser) Rune(i int) rune {
if i > 0 || p.keepRanges {
return p.parseRune(p.getField(i))
}
return p.rangeStart
}
// Runes interprets and returns field i as a sequence of runes.
func (p *Parser) Runes(i int) (runes []rune) {
add := func(s string) {
if s = strings.TrimSpace(s); len(s) > 0 {
runes = append(runes, p.parseRune(s))
}
}
for b := p.getField(i); ; {
i := strings.IndexByte(b, ' ')
if i == -1 {
add(b)
break
}
add(b[:i])
b = b[i+1:]
}
return
}
var (
errIncorrectLegacyRange = errors.New("ucd: unmatched <* First>")
// reRange matches one line of a legacy rune range.
reRange = regexp.MustCompile("^([0-9A-F]*);<([^,]*), ([^>]*)>(.*)$")
)
// Range parses and returns field i as a rune range. A range is inclusive at
// both ends. If the field only has one rune, first and last will be identical.
// It supports the legacy format for ranges used in UnicodeData.txt.
func (p *Parser) Range(i int) (first, last rune) {
if !p.keepRanges {
return p.rangeStart, p.rangeStart
}
return p.getRange(i)
}
func (p *Parser) getRange(i int) (first, last rune) {
b := p.getField(i)
if k := strings.Index(b, ".."); k != -1 {
return p.parseRune(b[:k]), p.parseRune(b[k+2:])
}
// The first field may not be a rune, in which case we may ignore any error
// and set the range as 0..0.
x, err := parseRune(b)
if err != nil {
// Disable range parsing henceforth. This ensures that an error will be
// returned if the user subsequently will try to parse this field as
// a Rune.
p.keepRanges = true
}
// Special case for UnicodeData that was retained for backwards compatibility.
if i == 0 && len(p.field) > 1 && strings.HasSuffix(p.field[1], "First>") {
if p.parsedRange {
return p.rangeStart, p.rangeEnd
}
mf := reRange.FindStringSubmatch(p.scanner.Text())
p.line++
if mf == nil || !p.scanner.Scan() {
p.setError(errIncorrectLegacyRange, "")
return x, x
}
// Using Bytes would be more efficient here, but Text is a lot easier
// and this is not a frequent case.
ml := reRange.FindStringSubmatch(p.scanner.Text())
if ml == nil || mf[2] != ml[2] || ml[3] != "Last" || mf[4] != ml[4] {
p.setError(errIncorrectLegacyRange, "")
return x, x
}
p.rangeStart, p.rangeEnd = x, p.parseRune(p.scanner.Text()[:len(ml[1])])
p.parsedRange = true
return p.rangeStart, p.rangeEnd
}
return x, x
}
// bools recognizes all valid UCD boolean values.
var bools = map[string]bool{
"": false,
"N": false,
"No": false,
"F": false,
"False": false,
"Y": true,
"Yes": true,
"T": true,
"True": true,
}
// Bool parses and returns field i as a boolean value.
func (p *Parser) Bool(i int) bool {
f := p.getField(i)
for s, v := range bools {
if f == s {
return v
}
}
p.setError(strconv.ErrSyntax, "error parsing bool")
return false
}
// Int parses and returns field i as an integer value.
func (p *Parser) Int(i int) int {
x, err := strconv.ParseInt(string(p.getField(i)), 10, 64)
p.setError(err, "error parsing int")
return int(x)
}
// Uint parses and returns field i as an unsigned integer value.
func (p *Parser) Uint(i int) uint {
x, err := strconv.ParseUint(string(p.getField(i)), 10, 64)
p.setError(err, "error parsing uint")
return uint(x)
}
// Float parses and returns field i as a decimal value.
func (p *Parser) Float(i int) float64 {
x, err := strconv.ParseFloat(string(p.getField(i)), 64)
p.setError(err, "error parsing float")
return x
}
// String parses and returns field i as a string value.
func (p *Parser) String(i int) string {
return string(p.getField(i))
}
// Strings parses and returns field i as a space-separated list of strings.
func (p *Parser) Strings(i int) []string {
ss := strings.Split(string(p.getField(i)), " ")
for i, s := range ss {
ss[i] = strings.TrimSpace(s)
}
return ss
}
// Comment returns the comments for the current line.
func (p *Parser) Comment() string {
return string(p.comment)
}
var errUndefinedEnum = errors.New("ucd: undefined enum value")
// Enum interprets and returns field i as a value that must be one of the values
// in enum.
func (p *Parser) Enum(i int, enum ...string) string {
f := p.getField(i)
for _, s := range enum {
if f == s {
return s
}
}
p.setError(errUndefinedEnum, "error parsing enum")
return ""
}

305
vendor/golang.org/x/text/language/gen.go generated vendored
View File

@ -1,305 +0,0 @@
// 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.
// +build ignore
// Language tag table generator.
// Data read from the web.
package main
import (
"flag"
"fmt"
"io"
"log"
"sort"
"strconv"
"strings"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/language"
"golang.org/x/text/unicode/cldr"
)
var (
test = flag.Bool("test",
false,
"test existing tables; can be used to compare web data with package data.")
outputFile = flag.String("output",
"tables.go",
"output file for generated tables")
)
func main() {
gen.Init()
w := gen.NewCodeWriter()
defer w.WriteGoFile("tables.go", "language")
b := newBuilder(w)
gen.WriteCLDRVersion(w)
b.writeConstants()
b.writeMatchData()
}
type builder struct {
w *gen.CodeWriter
hw io.Writer // MultiWriter for w and w.Hash
data *cldr.CLDR
supp *cldr.SupplementalData
}
func (b *builder) langIndex(s string) uint16 {
return uint16(language.MustParseBase(s))
}
func (b *builder) regionIndex(s string) int {
return int(language.MustParseRegion(s))
}
func (b *builder) scriptIndex(s string) int {
return int(language.MustParseScript(s))
}
func newBuilder(w *gen.CodeWriter) *builder {
r := gen.OpenCLDRCoreZip()
defer r.Close()
d := &cldr.Decoder{}
data, err := d.DecodeZip(r)
if err != nil {
log.Fatal(err)
}
b := builder{
w: w,
hw: io.MultiWriter(w, w.Hash),
data: data,
supp: data.Supplemental(),
}
return &b
}
// writeConsts computes f(v) for all v in values and writes the results
// as constants named _v to a single constant block.
func (b *builder) writeConsts(f func(string) int, values ...string) {
fmt.Fprintln(b.w, "const (")
for _, v := range values {
fmt.Fprintf(b.w, "\t_%s = %v\n", v, f(v))
}
fmt.Fprintln(b.w, ")")
}
// TODO: region inclusion data will probably not be use used in future matchers.
var langConsts = []string{
"de", "en", "fr", "it", "mo", "no", "nb", "pt", "sh", "mul", "und",
}
var scriptConsts = []string{
"Latn", "Hani", "Hans", "Hant", "Qaaa", "Qaai", "Qabx", "Zinh", "Zyyy",
"Zzzz",
}
var regionConsts = []string{
"001", "419", "BR", "CA", "ES", "GB", "MD", "PT", "UK", "US",
"ZZ", "XA", "XC", "XK", // Unofficial tag for Kosovo.
}
func (b *builder) writeConstants() {
b.writeConsts(func(s string) int { return int(b.langIndex(s)) }, langConsts...)
b.writeConsts(b.regionIndex, regionConsts...)
b.writeConsts(b.scriptIndex, scriptConsts...)
}
type mutualIntelligibility struct {
want, have uint16
distance uint8
oneway bool
}
type scriptIntelligibility struct {
wantLang, haveLang uint16
wantScript, haveScript uint8
distance uint8
// Always oneway
}
type regionIntelligibility struct {
lang uint16 // compact language id
script uint8 // 0 means any
group uint8 // 0 means any; if bit 7 is set it means inverse
distance uint8
// Always twoway.
}
// writeMatchData writes tables with languages and scripts for which there is
// mutual intelligibility. The data is based on CLDR's languageMatching data.
// Note that we use a different algorithm than the one defined by CLDR and that
// we slightly modify the data. For example, we convert scores to confidence levels.
// We also drop all region-related data as we use a different algorithm to
// determine region equivalence.
func (b *builder) writeMatchData() {
lm := b.supp.LanguageMatching.LanguageMatches
cldr.MakeSlice(&lm).SelectAnyOf("type", "written_new")
regionHierarchy := map[string][]string{}
for _, g := range b.supp.TerritoryContainment.Group {
regions := strings.Split(g.Contains, " ")
regionHierarchy[g.Type] = append(regionHierarchy[g.Type], regions...)
}
regionToGroups := make([]uint8, language.NumRegions)
idToIndex := map[string]uint8{}
for i, mv := range lm[0].MatchVariable {
if i > 6 {
log.Fatalf("Too many groups: %d", i)
}
idToIndex[mv.Id] = uint8(i + 1)
// TODO: also handle '-'
for _, r := range strings.Split(mv.Value, "+") {
todo := []string{r}
for k := 0; k < len(todo); k++ {
r := todo[k]
regionToGroups[b.regionIndex(r)] |= 1 << uint8(i)
todo = append(todo, regionHierarchy[r]...)
}
}
}
b.w.WriteVar("regionToGroups", regionToGroups)
// maps language id to in- and out-of-group region.
paradigmLocales := [][3]uint16{}
locales := strings.Split(lm[0].ParadigmLocales[0].Locales, " ")
for i := 0; i < len(locales); i += 2 {
x := [3]uint16{}
for j := 0; j < 2; j++ {
pc := strings.SplitN(locales[i+j], "-", 2)
x[0] = b.langIndex(pc[0])
if len(pc) == 2 {
x[1+j] = uint16(b.regionIndex(pc[1]))
}
}
paradigmLocales = append(paradigmLocales, x)
}
b.w.WriteVar("paradigmLocales", paradigmLocales)
b.w.WriteType(mutualIntelligibility{})
b.w.WriteType(scriptIntelligibility{})
b.w.WriteType(regionIntelligibility{})
matchLang := []mutualIntelligibility{}
matchScript := []scriptIntelligibility{}
matchRegion := []regionIntelligibility{}
// Convert the languageMatch entries in lists keyed by desired language.
for _, m := range lm[0].LanguageMatch {
// Different versions of CLDR use different separators.
desired := strings.Replace(m.Desired, "-", "_", -1)
supported := strings.Replace(m.Supported, "-", "_", -1)
d := strings.Split(desired, "_")
s := strings.Split(supported, "_")
if len(d) != len(s) {
log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
continue
}
distance, _ := strconv.ParseInt(m.Distance, 10, 8)
switch len(d) {
case 2:
if desired == supported && desired == "*_*" {
continue
}
// language-script pair.
matchScript = append(matchScript, scriptIntelligibility{
wantLang: uint16(b.langIndex(d[0])),
haveLang: uint16(b.langIndex(s[0])),
wantScript: uint8(b.scriptIndex(d[1])),
haveScript: uint8(b.scriptIndex(s[1])),
distance: uint8(distance),
})
if m.Oneway != "true" {
matchScript = append(matchScript, scriptIntelligibility{
wantLang: uint16(b.langIndex(s[0])),
haveLang: uint16(b.langIndex(d[0])),
wantScript: uint8(b.scriptIndex(s[1])),
haveScript: uint8(b.scriptIndex(d[1])),
distance: uint8(distance),
})
}
case 1:
if desired == supported && desired == "*" {
continue
}
if distance == 1 {
// nb == no is already handled by macro mapping. Check there
// really is only this case.
if d[0] != "no" || s[0] != "nb" {
log.Fatalf("unhandled equivalence %s == %s", s[0], d[0])
}
continue
}
// TODO: consider dropping oneway field and just doubling the entry.
matchLang = append(matchLang, mutualIntelligibility{
want: uint16(b.langIndex(d[0])),
have: uint16(b.langIndex(s[0])),
distance: uint8(distance),
oneway: m.Oneway == "true",
})
case 3:
if desired == supported && desired == "*_*_*" {
continue
}
if desired != supported {
// This is now supported by CLDR, but only one case, which
// should already be covered by paradigm locales. For instance,
// test case "und, en, en-GU, en-IN, en-GB ; en-ZA ; en-GB" in
// testdata/CLDRLocaleMatcherTest.txt tests this.
if supported != "en_*_GB" {
log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
}
continue
}
ri := regionIntelligibility{
lang: b.langIndex(d[0]),
distance: uint8(distance),
}
if d[1] != "*" {
ri.script = uint8(b.scriptIndex(d[1]))
}
switch {
case d[2] == "*":
ri.group = 0x80 // not contained in anything
case strings.HasPrefix(d[2], "$!"):
ri.group = 0x80
d[2] = "$" + d[2][len("$!"):]
fallthrough
case strings.HasPrefix(d[2], "$"):
ri.group |= idToIndex[d[2]]
}
matchRegion = append(matchRegion, ri)
default:
log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
}
}
sort.SliceStable(matchLang, func(i, j int) bool {
return matchLang[i].distance < matchLang[j].distance
})
b.w.WriteComment(`
matchLang holds pairs of langIDs of base languages that are typically
mutually intelligible. Each pair is associated with a confidence and
whether the intelligibility goes one or both ways.`)
b.w.WriteVar("matchLang", matchLang)
b.w.WriteComment(`
matchScript holds pairs of scriptIDs where readers of one script
can typically also read the other. Each is associated with a confidence.`)
sort.SliceStable(matchScript, func(i, j int) bool {
return matchScript[i].distance < matchScript[j].distance
})
b.w.WriteVar("matchScript", matchScript)
sort.SliceStable(matchRegion, func(i, j int) bool {
return matchRegion[i].distance < matchRegion[j].distance
})
b.w.WriteVar("matchRegion", matchRegion)
}

133
vendor/golang.org/x/text/unicode/bidi/gen.go generated vendored
View File

@ -1,133 +0,0 @@
// Copyright 2015 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
package main
import (
"flag"
"log"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/triegen"
"golang.org/x/text/internal/ucd"
)
var outputFile = flag.String("out", "tables.go", "output file")
func main() {
gen.Init()
gen.Repackage("gen_trieval.go", "trieval.go", "bidi")
gen.Repackage("gen_ranges.go", "ranges_test.go", "bidi")
genTables()
}
// bidiClass names and codes taken from class "bc" in
// https://www.unicode.org/Public/8.0.0/ucd/PropertyValueAliases.txt
var bidiClass = map[string]Class{
"AL": AL, // ArabicLetter
"AN": AN, // ArabicNumber
"B": B, // ParagraphSeparator
"BN": BN, // BoundaryNeutral
"CS": CS, // CommonSeparator
"EN": EN, // EuropeanNumber
"ES": ES, // EuropeanSeparator
"ET": ET, // EuropeanTerminator
"L": L, // LeftToRight
"NSM": NSM, // NonspacingMark
"ON": ON, // OtherNeutral
"R": R, // RightToLeft
"S": S, // SegmentSeparator
"WS": WS, // WhiteSpace
"FSI": Control,
"PDF": Control,
"PDI": Control,
"LRE": Control,
"LRI": Control,
"LRO": Control,
"RLE": Control,
"RLI": Control,
"RLO": Control,
}
func genTables() {
if numClass > 0x0F {
log.Fatalf("Too many Class constants (%#x > 0x0F).", numClass)
}
w := gen.NewCodeWriter()
defer w.WriteVersionedGoFile(*outputFile, "bidi")
gen.WriteUnicodeVersion(w)
t := triegen.NewTrie("bidi")
// Build data about bracket mapping. These bits need to be or-ed with
// any other bits.
orMask := map[rune]uint64{}
xorMap := map[rune]int{}
xorMasks := []rune{0} // First value is no-op.
ucd.Parse(gen.OpenUCDFile("BidiBrackets.txt"), func(p *ucd.Parser) {
r1 := p.Rune(0)
r2 := p.Rune(1)
xor := r1 ^ r2
if _, ok := xorMap[xor]; !ok {
xorMap[xor] = len(xorMasks)
xorMasks = append(xorMasks, xor)
}
entry := uint64(xorMap[xor]) << xorMaskShift
switch p.String(2) {
case "o":
entry |= openMask
case "c", "n":
default:
log.Fatalf("Unknown bracket class %q.", p.String(2))
}
orMask[r1] = entry
})
w.WriteComment(`
xorMasks contains masks to be xor-ed with brackets to get the reverse
version.`)
w.WriteVar("xorMasks", xorMasks)
done := map[rune]bool{}
insert := func(r rune, c Class) {
if !done[r] {
t.Insert(r, orMask[r]|uint64(c))
done[r] = true
}
}
// Insert the derived BiDi properties.
ucd.Parse(gen.OpenUCDFile("extracted/DerivedBidiClass.txt"), func(p *ucd.Parser) {
r := p.Rune(0)
class, ok := bidiClass[p.String(1)]
if !ok {
log.Fatalf("%U: Unknown BiDi class %q", r, p.String(1))
}
insert(r, class)
})
visitDefaults(insert)
// TODO: use sparse blocks. This would reduce table size considerably
// from the looks of it.
sz, err := t.Gen(w)
if err != nil {
log.Fatal(err)
}
w.Size += sz
}
// dummy values to make methods in gen_common compile. The real versions
// will be generated by this file to tables.go.
var (
xorMasks []rune
)

57
vendor/golang.org/x/text/unicode/bidi/gen_ranges.go generated vendored
View File

@ -1,57 +0,0 @@
// Copyright 2015 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
package main
import (
"unicode"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/ucd"
"golang.org/x/text/unicode/rangetable"
)
// These tables are hand-extracted from:
// https://www.unicode.org/Public/8.0.0/ucd/extracted/DerivedBidiClass.txt
func visitDefaults(fn func(r rune, c Class)) {
// first write default values for ranges listed above.
visitRunes(fn, AL, []rune{
0x0600, 0x07BF, // Arabic
0x08A0, 0x08FF, // Arabic Extended-A
0xFB50, 0xFDCF, // Arabic Presentation Forms
0xFDF0, 0xFDFF,
0xFE70, 0xFEFF,
0x0001EE00, 0x0001EEFF, // Arabic Mathematical Alpha Symbols
})
visitRunes(fn, R, []rune{
0x0590, 0x05FF, // Hebrew
0x07C0, 0x089F, // Nko et al.
0xFB1D, 0xFB4F,
0x00010800, 0x00010FFF, // Cypriot Syllabary et. al.
0x0001E800, 0x0001EDFF,
0x0001EF00, 0x0001EFFF,
})
visitRunes(fn, ET, []rune{ // European Terminator
0x20A0, 0x20Cf, // Currency symbols
})
rangetable.Visit(unicode.Noncharacter_Code_Point, func(r rune) {
fn(r, BN) // Boundary Neutral
})
ucd.Parse(gen.OpenUCDFile("DerivedCoreProperties.txt"), func(p *ucd.Parser) {
if p.String(1) == "Default_Ignorable_Code_Point" {
fn(p.Rune(0), BN) // Boundary Neutral
}
})
}
func visitRunes(fn func(r rune, c Class), c Class, runes []rune) {
for i := 0; i < len(runes); i += 2 {
lo, hi := runes[i], runes[i+1]
for j := lo; j <= hi; j++ {
fn(j, c)
}
}
}

64
vendor/golang.org/x/text/unicode/bidi/gen_trieval.go generated vendored
View File

@ -1,64 +0,0 @@
// Copyright 2015 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
package main
// Class is the Unicode BiDi class. Each rune has a single class.
type Class uint
const (
L Class = iota // LeftToRight
R // RightToLeft
EN // EuropeanNumber
ES // EuropeanSeparator
ET // EuropeanTerminator
AN // ArabicNumber
CS // CommonSeparator
B // ParagraphSeparator
S // SegmentSeparator
WS // WhiteSpace
ON // OtherNeutral
BN // BoundaryNeutral
NSM // NonspacingMark
AL // ArabicLetter
Control // Control LRO - PDI
numClass
LRO // LeftToRightOverride
RLO // RightToLeftOverride
LRE // LeftToRightEmbedding
RLE // RightToLeftEmbedding
PDF // PopDirectionalFormat
LRI // LeftToRightIsolate
RLI // RightToLeftIsolate
FSI // FirstStrongIsolate
PDI // PopDirectionalIsolate
unknownClass = ^Class(0)
)
var controlToClass = map[rune]Class{
0x202D: LRO, // LeftToRightOverride,
0x202E: RLO, // RightToLeftOverride,
0x202A: LRE, // LeftToRightEmbedding,
0x202B: RLE, // RightToLeftEmbedding,
0x202C: PDF, // PopDirectionalFormat,
0x2066: LRI, // LeftToRightIsolate,
0x2067: RLI, // RightToLeftIsolate,
0x2068: FSI, // FirstStrongIsolate,
0x2069: PDI, // PopDirectionalIsolate,
}
// A trie entry has the following bits:
// 7..5 XOR mask for brackets
// 4 1: Bracket open, 0: Bracket close
// 3..0 Class type
const (
openMask = 0x10
xorMaskShift = 5
)

105
vendor/golang.org/x/text/unicode/cldr/base.go generated vendored
View File

@ -1,105 +0,0 @@
// 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 cldr
import (
"encoding/xml"
"regexp"
"strconv"
)
// Elem is implemented by every XML element.
type Elem interface {
setEnclosing(Elem)
setName(string)
enclosing() Elem
GetCommon() *Common
}
type hidden struct {
CharData string `xml:",chardata"`
Alias *struct {
Common
Source string `xml:"source,attr"`
Path string `xml:"path,attr"`
} `xml:"alias"`
Def *struct {
Common
Choice string `xml:"choice,attr,omitempty"`
Type string `xml:"type,attr,omitempty"`
} `xml:"default"`
}
// Common holds several of the most common attributes and sub elements
// of an XML element.
type Common struct {
XMLName xml.Name
name string
enclElem Elem
Type string `xml:"type,attr,omitempty"`
Reference string `xml:"reference,attr,omitempty"`
Alt string `xml:"alt,attr,omitempty"`
ValidSubLocales string `xml:"validSubLocales,attr,omitempty"`
Draft string `xml:"draft,attr,omitempty"`
hidden
}
// Default returns the default type to select from the enclosed list
// or "" if no default value is specified.
func (e *Common) Default() string {
if e.Def == nil {
return ""
}
if e.Def.Choice != "" {
return e.Def.Choice
} else if e.Def.Type != "" {
// Type is still used by the default element in collation.
return e.Def.Type
}
return ""
}
// Element returns the XML element name.
func (e *Common) Element() string {
return e.name
}
// GetCommon returns e. It is provided such that Common implements Elem.
func (e *Common) GetCommon() *Common {
return e
}
// Data returns the character data accumulated for this element.
func (e *Common) Data() string {
e.CharData = charRe.ReplaceAllStringFunc(e.CharData, replaceUnicode)
return e.CharData
}
func (e *Common) setName(s string) {
e.name = s
}
func (e *Common) enclosing() Elem {
return e.enclElem
}
func (e *Common) setEnclosing(en Elem) {
e.enclElem = en
}
// Escape characters that can be escaped without further escaping the string.
var charRe = regexp.MustCompile(`&#x[0-9a-fA-F]*;|\\u[0-9a-fA-F]{4}|\\U[0-9a-fA-F]{8}|\\x[0-9a-fA-F]{2}|\\[0-7]{3}|\\[abtnvfr]`)
// replaceUnicode converts hexadecimal Unicode codepoint notations to a one-rune string.
// It assumes the input string is correctly formatted.
func replaceUnicode(s string) string {
if s[1] == '#' {
r, _ := strconv.ParseInt(s[3:len(s)-1], 16, 32)
return string(r)
}
r, _, _, _ := strconv.UnquoteChar(s, 0)
return string(r)
}

137
vendor/golang.org/x/text/unicode/cldr/cldr.go generated vendored
View File

@ -1,137 +0,0 @@
// 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.
//go:generate go run makexml.go -output xml.go
// Package cldr provides a parser for LDML and related XML formats.
//
// This package is intended to be used by the table generation tools for the
// various packages in x/text and is not internal for historical reasons.
//
// As the XML types are generated from the CLDR DTD, and as the CLDR standard is
// periodically amended, this package may change considerably over time. This
// mostly means that data may appear and disappear between versions. That is,
// old code should keep compiling for newer versions, but data may have moved or
// changed. CLDR version 22 is the first version supported by this package.
// Older versions may not work.
package cldr // import "golang.org/x/text/unicode/cldr"
import (
"fmt"
"sort"
)
// CLDR provides access to parsed data of the Unicode Common Locale Data Repository.
type CLDR struct {
parent map[string][]string
locale map[string]*LDML
resolved map[string]*LDML
bcp47 *LDMLBCP47
supp *SupplementalData
}
func makeCLDR() *CLDR {
return &CLDR{
parent: make(map[string][]string),
locale: make(map[string]*LDML),
resolved: make(map[string]*LDML),
bcp47: &LDMLBCP47{},
supp: &SupplementalData{},
}
}
// BCP47 returns the parsed BCP47 LDML data. If no such data was parsed, nil is returned.
func (cldr *CLDR) BCP47() *LDMLBCP47 {
return nil
}
// Draft indicates the draft level of an element.
type Draft int
const (
Approved Draft = iota
Contributed
Provisional
Unconfirmed
)
var drafts = []string{"unconfirmed", "provisional", "contributed", "approved", ""}
// ParseDraft returns the Draft value corresponding to the given string. The
// empty string corresponds to Approved.
func ParseDraft(level string) (Draft, error) {
if level == "" {
return Approved, nil
}
for i, s := range drafts {
if level == s {
return Unconfirmed - Draft(i), nil
}
}
return Approved, fmt.Errorf("cldr: unknown draft level %q", level)
}
func (d Draft) String() string {
return drafts[len(drafts)-1-int(d)]
}
// SetDraftLevel sets which draft levels to include in the evaluated LDML.
// Any draft element for which the draft level is higher than lev will be excluded.
// If multiple draft levels are available for a single element, the one with the
// lowest draft level will be selected, unless preferDraft is true, in which case
// the highest draft will be chosen.
// It is assumed that the underlying LDML is canonicalized.
func (cldr *CLDR) SetDraftLevel(lev Draft, preferDraft bool) {
// TODO: implement
cldr.resolved = make(map[string]*LDML)
}
// RawLDML returns the LDML XML for id in unresolved form.
// id must be one of the strings returned by Locales.
func (cldr *CLDR) RawLDML(loc string) *LDML {
return cldr.locale[loc]
}
// LDML returns the fully resolved LDML XML for loc, which must be one of
// the strings returned by Locales.
//
// Deprecated: Use RawLDML and implement inheritance manually or using the
// internal cldrtree package.
// Inheritance has changed quite a bit since the onset of this package and in
// practice data often represented in a way where knowledge of how it was
// inherited is relevant.
func (cldr *CLDR) LDML(loc string) (*LDML, error) {
return cldr.resolve(loc)
}
// Supplemental returns the parsed supplemental data. If no such data was parsed,
// nil is returned.
func (cldr *CLDR) Supplemental() *SupplementalData {
return cldr.supp
}
// Locales returns the locales for which there exist files.
// Valid sublocales for which there is no file are not included.
// The root locale is always sorted first.
func (cldr *CLDR) Locales() []string {
loc := []string{"root"}
hasRoot := false
for l, _ := range cldr.locale {
if l == "root" {
hasRoot = true
continue
}
loc = append(loc, l)
}
sort.Strings(loc[1:])
if !hasRoot {
return loc[1:]
}
return loc
}
// Get fills in the fields of x based on the XPath path.
func Get(e Elem, path string) (res Elem, err error) {
return walkXPath(e, path)
}

359
vendor/golang.org/x/text/unicode/cldr/collate.go generated vendored
View File

@ -1,359 +0,0 @@
// 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 cldr
import (
"bufio"
"encoding/xml"
"errors"
"fmt"
"strconv"
"strings"
"unicode"
"unicode/utf8"
)
// RuleProcessor can be passed to Collator's Process method, which
// parses the rules and calls the respective method for each rule found.
type RuleProcessor interface {
Reset(anchor string, before int) error
Insert(level int, str, context, extend string) error
Index(id string)
}
const (
// cldrIndex is a Unicode-reserved sentinel value used to mark the start
// of a grouping within an index.
// We ignore any rule that starts with this rune.
// See https://unicode.org/reports/tr35/#Collation_Elements for details.
cldrIndex = "\uFDD0"
// specialAnchor is the format in which to represent logical reset positions,
// such as "first tertiary ignorable".
specialAnchor = "<%s/>"
)
// Process parses the rules for the tailorings of this collation
// and calls the respective methods of p for each rule found.
func (c Collation) Process(p RuleProcessor) (err error) {
if len(c.Cr) > 0 {
if len(c.Cr) > 1 {
return fmt.Errorf("multiple cr elements, want 0 or 1")
}
return processRules(p, c.Cr[0].Data())
}
if c.Rules.Any != nil {
return c.processXML(p)
}
return errors.New("no tailoring data")
}
// processRules parses rules in the Collation Rule Syntax defined in
// https://www.unicode.org/reports/tr35/tr35-collation.html#Collation_Tailorings.
func processRules(p RuleProcessor, s string) (err error) {
chk := func(s string, e error) string {
if err == nil {
err = e
}
return s
}
i := 0 // Save the line number for use after the loop.
scanner := bufio.NewScanner(strings.NewReader(s))
for ; scanner.Scan() && err == nil; i++ {
for s := skipSpace(scanner.Text()); s != "" && s[0] != '#'; s = skipSpace(s) {
level := 5
var ch byte
switch ch, s = s[0], s[1:]; ch {
case '&': // followed by <anchor> or '[' <key> ']'
if s = skipSpace(s); consume(&s, '[') {
s = chk(parseSpecialAnchor(p, s))
} else {
s = chk(parseAnchor(p, 0, s))
}
case '<': // sort relation '<'{1,4}, optionally followed by '*'.
for level = 1; consume(&s, '<'); level++ {
}
if level > 4 {
err = fmt.Errorf("level %d > 4", level)
}
fallthrough
case '=': // identity relation, optionally followed by *.
if consume(&s, '*') {
s = chk(parseSequence(p, level, s))
} else {
s = chk(parseOrder(p, level, s))
}
default:
chk("", fmt.Errorf("illegal operator %q", ch))
break
}
}
}
if chk("", scanner.Err()); err != nil {
return fmt.Errorf("%d: %v", i, err)
}
return nil
}
// parseSpecialAnchor parses the anchor syntax which is either of the form
// ['before' <level>] <anchor>
// or
// [<label>]
// The starting should already be consumed.
func parseSpecialAnchor(p RuleProcessor, s string) (tail string, err error) {
i := strings.IndexByte(s, ']')
if i == -1 {
return "", errors.New("unmatched bracket")
}
a := strings.TrimSpace(s[:i])
s = s[i+1:]
if strings.HasPrefix(a, "before ") {
l, err := strconv.ParseUint(skipSpace(a[len("before "):]), 10, 3)
if err != nil {
return s, err
}
return parseAnchor(p, int(l), s)
}
return s, p.Reset(fmt.Sprintf(specialAnchor, a), 0)
}
func parseAnchor(p RuleProcessor, level int, s string) (tail string, err error) {
anchor, s, err := scanString(s)
if err != nil {
return s, err
}
return s, p.Reset(anchor, level)
}
func parseOrder(p RuleProcessor, level int, s string) (tail string, err error) {
var value, context, extend string
if value, s, err = scanString(s); err != nil {
return s, err
}
if strings.HasPrefix(value, cldrIndex) {
p.Index(value[len(cldrIndex):])
return
}
if consume(&s, '|') {
if context, s, err = scanString(s); err != nil {
return s, errors.New("missing string after context")
}
}
if consume(&s, '/') {
if extend, s, err = scanString(s); err != nil {
return s, errors.New("missing string after extension")
}
}
return s, p.Insert(level, value, context, extend)
}
// scanString scans a single input string.
func scanString(s string) (str, tail string, err error) {
if s = skipSpace(s); s == "" {
return s, s, errors.New("missing string")
}
buf := [16]byte{} // small but enough to hold most cases.
value := buf[:0]
for s != "" {
if consume(&s, '\'') {
i := strings.IndexByte(s, '\'')
if i == -1 {
return "", "", errors.New(`unmatched single quote`)
}
if i == 0 {
value = append(value, '\'')
} else {
value = append(value, s[:i]...)
}
s = s[i+1:]
continue
}
r, sz := utf8.DecodeRuneInString(s)
if unicode.IsSpace(r) || strings.ContainsRune("&<=#", r) {
break
}
value = append(value, s[:sz]...)
s = s[sz:]
}
return string(value), skipSpace(s), nil
}
func parseSequence(p RuleProcessor, level int, s string) (tail string, err error) {
if s = skipSpace(s); s == "" {
return s, errors.New("empty sequence")
}
last := rune(0)
for s != "" {
r, sz := utf8.DecodeRuneInString(s)
s = s[sz:]
if r == '-' {
// We have a range. The first element was already written.
if last == 0 {
return s, errors.New("range without starter value")
}
r, sz = utf8.DecodeRuneInString(s)
s = s[sz:]
if r == utf8.RuneError || r < last {
return s, fmt.Errorf("invalid range %q-%q", last, r)
}
for i := last + 1; i <= r; i++ {
if err := p.Insert(level, string(i), "", ""); err != nil {
return s, err
}
}
last = 0
continue
}
if unicode.IsSpace(r) || unicode.IsPunct(r) {
break
}
// normal case
if err := p.Insert(level, string(r), "", ""); err != nil {
return s, err
}
last = r
}
return s, nil
}
func skipSpace(s string) string {
return strings.TrimLeftFunc(s, unicode.IsSpace)
}
// consumes returns whether the next byte is ch. If so, it gobbles it by
// updating s.
func consume(s *string, ch byte) (ok bool) {
if *s == "" || (*s)[0] != ch {
return false
}
*s = (*s)[1:]
return true
}
// The following code parses Collation rules of CLDR version 24 and before.
var lmap = map[byte]int{
'p': 1,
's': 2,
't': 3,
'i': 5,
}
type rulesElem struct {
Rules struct {
Common
Any []*struct {
XMLName xml.Name
rule
} `xml:",any"`
} `xml:"rules"`
}
type rule struct {
Value string `xml:",chardata"`
Before string `xml:"before,attr"`
Any []*struct {
XMLName xml.Name
rule
} `xml:",any"`
}
var emptyValueError = errors.New("cldr: empty rule value")
func (r *rule) value() (string, error) {
// Convert hexadecimal Unicode codepoint notation to a string.
s := charRe.ReplaceAllStringFunc(r.Value, replaceUnicode)
r.Value = s
if s == "" {
if len(r.Any) != 1 {
return "", emptyValueError
}
r.Value = fmt.Sprintf(specialAnchor, r.Any[0].XMLName.Local)
r.Any = nil
} else if len(r.Any) != 0 {
return "", fmt.Errorf("cldr: XML elements found in collation rule: %v", r.Any)
}
return r.Value, nil
}
func (r rule) process(p RuleProcessor, name, context, extend string) error {
v, err := r.value()
if err != nil {
return err
}
switch name {
case "p", "s", "t", "i":
if strings.HasPrefix(v, cldrIndex) {
p.Index(v[len(cldrIndex):])
return nil
}
if err := p.Insert(lmap[name[0]], v, context, extend); err != nil {
return err
}
case "pc", "sc", "tc", "ic":
level := lmap[name[0]]
for _, s := range v {
if err := p.Insert(level, string(s), context, extend); err != nil {
return err
}
}
default:
return fmt.Errorf("cldr: unsupported tag: %q", name)
}
return nil
}
// processXML parses the format of CLDR versions 24 and older.
func (c Collation) processXML(p RuleProcessor) (err error) {
// Collation is generated and defined in xml.go.
var v string
for _, r := range c.Rules.Any {
switch r.XMLName.Local {
case "reset":
level := 0
switch r.Before {
case "primary", "1":
level = 1
case "secondary", "2":
level = 2
case "tertiary", "3":
level = 3
case "":
default:
return fmt.Errorf("cldr: unknown level %q", r.Before)
}
v, err = r.value()
if err == nil {
err = p.Reset(v, level)
}
case "x":
var context, extend string
for _, r1 := range r.Any {
v, err = r1.value()
switch r1.XMLName.Local {
case "context":
context = v
case "extend":
extend = v
}
}
for _, r1 := range r.Any {
if t := r1.XMLName.Local; t == "context" || t == "extend" {
continue
}
r1.rule.process(p, r1.XMLName.Local, context, extend)
}
default:
err = r.rule.process(p, r.XMLName.Local, "", "")
}
if err != nil {
return err
}
}
return nil
}

172
vendor/golang.org/x/text/unicode/cldr/decode.go generated vendored
View File

@ -1,172 +0,0 @@
// 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 cldr
import (
"archive/zip"
"bytes"
"encoding/xml"
"fmt"
"io"
"io/ioutil"
"log"
"os"
"path/filepath"
"regexp"
)
// A Decoder loads an archive of CLDR data.
type Decoder struct {
dirFilter []string
sectionFilter []string
loader Loader
cldr *CLDR
curLocale string
}
// SetSectionFilter takes a list top-level LDML element names to which
// evaluation of LDML should be limited. It automatically calls SetDirFilter.
func (d *Decoder) SetSectionFilter(filter ...string) {
d.sectionFilter = filter
// TODO: automatically set dir filter
}
// SetDirFilter limits the loading of LDML XML files of the specied directories.
// Note that sections may be split across directories differently for different CLDR versions.
// For more robust code, use SetSectionFilter.
func (d *Decoder) SetDirFilter(dir ...string) {
d.dirFilter = dir
}
// A Loader provides access to the files of a CLDR archive.
type Loader interface {
Len() int
Path(i int) string
Reader(i int) (io.ReadCloser, error)
}
var fileRe = regexp.MustCompile(`.*[/\\](.*)[/\\](.*)\.xml`)
// Decode loads and decodes the files represented by l.
func (d *Decoder) Decode(l Loader) (cldr *CLDR, err error) {
d.cldr = makeCLDR()
for i := 0; i < l.Len(); i++ {
fname := l.Path(i)
if m := fileRe.FindStringSubmatch(fname); m != nil {
if len(d.dirFilter) > 0 && !in(d.dirFilter, m[1]) {
continue
}
var r io.ReadCloser
if r, err = l.Reader(i); err == nil {
err = d.decode(m[1], m[2], r)
r.Close()
}
if err != nil {
return nil, err
}
}
}
d.cldr.finalize(d.sectionFilter)
return d.cldr, nil
}
func (d *Decoder) decode(dir, id string, r io.Reader) error {
var v interface{}
var l *LDML
cldr := d.cldr
switch {
case dir == "supplemental":
v = cldr.supp
case dir == "transforms":
return nil
case dir == "bcp47":
v = cldr.bcp47
case dir == "validity":
return nil
default:
ok := false
if v, ok = cldr.locale[id]; !ok {
l = &LDML{}
v, cldr.locale[id] = l, l
}
}
x := xml.NewDecoder(r)
if err := x.Decode(v); err != nil {
log.Printf("%s/%s: %v", dir, id, err)
return err
}
if l != nil {
if l.Identity == nil {
return fmt.Errorf("%s/%s: missing identity element", dir, id)
}
// TODO: verify when CLDR bug https://unicode.org/cldr/trac/ticket/8970
// is resolved.
// path := strings.Split(id, "_")
// if lang := l.Identity.Language.Type; lang != path[0] {
// return fmt.Errorf("%s/%s: language was %s; want %s", dir, id, lang, path[0])
// }
}
return nil
}
type pathLoader []string
func makePathLoader(path string) (pl pathLoader, err error) {
err = filepath.Walk(path, func(path string, _ os.FileInfo, err error) error {
pl = append(pl, path)
return err
})
return pl, err
}
func (pl pathLoader) Len() int {
return len(pl)
}
func (pl pathLoader) Path(i int) string {
return pl[i]
}
func (pl pathLoader) Reader(i int) (io.ReadCloser, error) {
return os.Open(pl[i])
}
// DecodePath loads CLDR data from the given path.
func (d *Decoder) DecodePath(path string) (cldr *CLDR, err error) {
loader, err := makePathLoader(path)
if err != nil {
return nil, err
}
return d.Decode(loader)
}
type zipLoader struct {
r *zip.Reader
}
func (zl zipLoader) Len() int {
return len(zl.r.File)
}
func (zl zipLoader) Path(i int) string {
return zl.r.File[i].Name
}
func (zl zipLoader) Reader(i int) (io.ReadCloser, error) {
return zl.r.File[i].Open()
}
// DecodeZip loads CLDR data from the zip archive for which r is the source.
func (d *Decoder) DecodeZip(r io.Reader) (cldr *CLDR, err error) {
buffer, err := ioutil.ReadAll(r)
if err != nil {
return nil, err
}
archive, err := zip.NewReader(bytes.NewReader(buffer), int64(len(buffer)))
if err != nil {
return nil, err
}
return d.Decode(zipLoader{archive})
}

400
vendor/golang.org/x/text/unicode/cldr/makexml.go generated vendored
View File

@ -1,400 +0,0 @@
// 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.
// +build ignore
// This tool generates types for the various XML formats of CLDR.
package main
import (
"archive/zip"
"bytes"
"encoding/xml"
"flag"
"fmt"
"io"
"io/ioutil"
"log"
"os"
"regexp"
"strings"
"golang.org/x/text/internal/gen"
)
var outputFile = flag.String("output", "xml.go", "output file name")
func main() {
flag.Parse()
r := gen.OpenCLDRCoreZip()
buffer, err := ioutil.ReadAll(r)
if err != nil {
log.Fatal("Could not read zip file")
}
r.Close()
z, err := zip.NewReader(bytes.NewReader(buffer), int64(len(buffer)))
if err != nil {
log.Fatalf("Could not read zip archive: %v", err)
}
var buf bytes.Buffer
version := gen.CLDRVersion()
for _, dtd := range files {
for _, f := range z.File {
if strings.HasSuffix(f.Name, dtd.file+".dtd") {
r, err := f.Open()
failOnError(err)
b := makeBuilder(&buf, dtd)
b.parseDTD(r)
b.resolve(b.index[dtd.top[0]])
b.write()
if b.version != "" && version != b.version {
println(f.Name)
log.Fatalf("main: inconsistent versions: found %s; want %s", b.version, version)
}
break
}
}
}
fmt.Fprintln(&buf, "// Version is the version of CLDR from which the XML definitions are generated.")
fmt.Fprintf(&buf, "const Version = %q\n", version)
gen.WriteGoFile(*outputFile, "cldr", buf.Bytes())
}
func failOnError(err error) {
if err != nil {
log.New(os.Stderr, "", log.Lshortfile).Output(2, err.Error())
os.Exit(1)
}
}
// configuration data per DTD type
type dtd struct {
file string // base file name
root string // Go name of the root XML element
top []string // create a different type for this section
skipElem []string // hard-coded or deprecated elements
skipAttr []string // attributes to exclude
predefined []string // hard-coded elements exist of the form <name>Elem
forceRepeat []string // elements to make slices despite DTD
}
var files = []dtd{
{
file: "ldmlBCP47",
root: "LDMLBCP47",
top: []string{"ldmlBCP47"},
skipElem: []string{
"cldrVersion", // deprecated, not used
},
},
{
file: "ldmlSupplemental",
root: "SupplementalData",
top: []string{"supplementalData"},
skipElem: []string{
"cldrVersion", // deprecated, not used
},
forceRepeat: []string{
"plurals", // data defined in plurals.xml and ordinals.xml
},
},
{
file: "ldml",
root: "LDML",
top: []string{
"ldml", "collation", "calendar", "timeZoneNames", "localeDisplayNames", "numbers",
},
skipElem: []string{
"cp", // not used anywhere
"special", // not used anywhere
"fallback", // deprecated, not used
"alias", // in Common
"default", // in Common
},
skipAttr: []string{
"hiraganaQuarternary", // typo in DTD, correct version included as well
},
predefined: []string{"rules"},
},
}
var comments = map[string]string{
"ldmlBCP47": `
// LDMLBCP47 holds information on allowable values for various variables in LDML.
`,
"supplementalData": `
// SupplementalData holds information relevant for internationalization
// and proper use of CLDR, but that is not contained in the locale hierarchy.
`,
"ldml": `
// LDML is the top-level type for locale-specific data.
`,
"collation": `
// Collation contains rules that specify a certain sort-order,
// as a tailoring of the root order.
// The parsed rules are obtained by passing a RuleProcessor to Collation's
// Process method.
`,
"calendar": `
// Calendar specifies the fields used for formatting and parsing dates and times.
// The month and quarter names are identified numerically, starting at 1.
// The day (of the week) names are identified with short strings, since there is
// no universally-accepted numeric designation.
`,
"dates": `
// Dates contains information regarding the format and parsing of dates and times.
`,
"localeDisplayNames": `
// LocaleDisplayNames specifies localized display names for scripts, languages,
// countries, currencies, and variants.
`,
"numbers": `
// Numbers supplies information for formatting and parsing numbers and currencies.
`,
}
type element struct {
name string // XML element name
category string // elements contained by this element
signature string // category + attrKey*
attr []*attribute // attributes supported by this element.
sub []struct { // parsed and evaluated sub elements of this element.
e *element
repeat bool // true if the element needs to be a slice
}
resolved bool // prevent multiple resolutions of this element.
}
type attribute struct {
name string
key string
list []string
tag string // Go tag
}
var (
reHead = regexp.MustCompile(` *(\w+) +([\w\-]+)`)
reAttr = regexp.MustCompile(` *(\w+) *(?:(\w+)|\(([\w\- \|]+)\)) *(?:#([A-Z]*) *(?:\"([\.\d+])\")?)? *("[\w\-:]*")?`)
reElem = regexp.MustCompile(`^ *(EMPTY|ANY|\(.*\)[\*\+\?]?) *$`)
reToken = regexp.MustCompile(`\w\-`)
)
// builder is used to read in the DTD files from CLDR and generate Go code
// to be used with the encoding/xml package.
type builder struct {
w io.Writer
index map[string]*element
elem []*element
info dtd
version string
}
func makeBuilder(w io.Writer, d dtd) builder {
return builder{
w: w,
index: make(map[string]*element),
elem: []*element{},
info: d,
}
}
// parseDTD parses a DTD file.
func (b *builder) parseDTD(r io.Reader) {
for d := xml.NewDecoder(r); ; {
t, err := d.Token()
if t == nil {
break
}
failOnError(err)
dir, ok := t.(xml.Directive)
if !ok {
continue
}
m := reHead.FindSubmatch(dir)
dir = dir[len(m[0]):]
ename := string(m[2])
el, elementFound := b.index[ename]
switch string(m[1]) {
case "ELEMENT":
if elementFound {
log.Fatal("parseDTD: duplicate entry for element %q", ename)
}
m := reElem.FindSubmatch(dir)
if m == nil {
log.Fatalf("parseDTD: invalid element %q", string(dir))
}
if len(m[0]) != len(dir) {
log.Fatal("parseDTD: invalid element %q", string(dir), len(dir), len(m[0]), string(m[0]))
}
s := string(m[1])
el = &element{
name: ename,
category: s,
}
b.index[ename] = el
case "ATTLIST":
if !elementFound {
log.Fatalf("parseDTD: unknown element %q", ename)
}
s := string(dir)
m := reAttr.FindStringSubmatch(s)
if m == nil {
log.Fatal(fmt.Errorf("parseDTD: invalid attribute %q", string(dir)))
}
if m[4] == "FIXED" {
b.version = m[5]
} else {
switch m[1] {
case "draft", "references", "alt", "validSubLocales", "standard" /* in Common */ :
case "type", "choice":
default:
el.attr = append(el.attr, &attribute{
name: m[1],
key: s,
list: reToken.FindAllString(m[3], -1),
})
el.signature = fmt.Sprintf("%s=%s+%s", el.signature, m[1], m[2])
}
}
}
}
}
var reCat = regexp.MustCompile(`[ ,\|]*(?:(\(|\)|\#?[\w_-]+)([\*\+\?]?))?`)
// resolve takes a parsed element and converts it into structured data
// that can be used to generate the XML code.
func (b *builder) resolve(e *element) {
if e.resolved {
return
}
b.elem = append(b.elem, e)
e.resolved = true
s := e.category
found := make(map[string]bool)
sequenceStart := []int{}
for len(s) > 0 {
m := reCat.FindStringSubmatch(s)
if m == nil {
log.Fatalf("%s: invalid category string %q", e.name, s)
}
repeat := m[2] == "*" || m[2] == "+" || in(b.info.forceRepeat, m[1])
switch m[1] {
case "":
case "(":
sequenceStart = append(sequenceStart, len(e.sub))
case ")":
if len(sequenceStart) == 0 {
log.Fatalf("%s: unmatched closing parenthesis", e.name)
}
for i := sequenceStart[len(sequenceStart)-1]; i < len(e.sub); i++ {
e.sub[i].repeat = e.sub[i].repeat || repeat
}
sequenceStart = sequenceStart[:len(sequenceStart)-1]
default:
if in(b.info.skipElem, m[1]) {
} else if sub, ok := b.index[m[1]]; ok {
if !found[sub.name] {
e.sub = append(e.sub, struct {
e *element
repeat bool
}{sub, repeat})
found[sub.name] = true
b.resolve(sub)
}
} else if m[1] == "#PCDATA" || m[1] == "ANY" {
} else if m[1] != "EMPTY" {
log.Fatalf("resolve:%s: element %q not found", e.name, m[1])
}
}
s = s[len(m[0]):]
}
}
// return true if s is contained in set.
func in(set []string, s string) bool {
for _, v := range set {
if v == s {
return true
}
}
return false
}
var repl = strings.NewReplacer("-", " ", "_", " ")
// title puts the first character or each character following '_' in title case and
// removes all occurrences of '_'.
func title(s string) string {
return strings.Replace(strings.Title(repl.Replace(s)), " ", "", -1)
}
// writeElem generates Go code for a single element, recursively.
func (b *builder) writeElem(tab int, e *element) {
p := func(f string, x ...interface{}) {
f = strings.Replace(f, "\n", "\n"+strings.Repeat("\t", tab), -1)
fmt.Fprintf(b.w, f, x...)
}
if len(e.sub) == 0 && len(e.attr) == 0 {
p("Common")
return
}
p("struct {")
tab++
p("\nCommon")
for _, attr := range e.attr {
if !in(b.info.skipAttr, attr.name) {
p("\n%s string `xml:\"%s,attr\"`", title(attr.name), attr.name)
}
}
for _, sub := range e.sub {
if in(b.info.predefined, sub.e.name) {
p("\n%sElem", sub.e.name)
continue
}
if in(b.info.skipElem, sub.e.name) {
continue
}
p("\n%s ", title(sub.e.name))
if sub.repeat {
p("[]")
}
p("*")
if in(b.info.top, sub.e.name) {
p(title(sub.e.name))
} else {
b.writeElem(tab, sub.e)
}
p(" `xml:\"%s\"`", sub.e.name)
}
tab--
p("\n}")
}
// write generates the Go XML code.
func (b *builder) write() {
for i, name := range b.info.top {
e := b.index[name]
if e != nil {
fmt.Fprintf(b.w, comments[name])
name := title(e.name)
if i == 0 {
name = b.info.root
}
fmt.Fprintf(b.w, "type %s ", name)
b.writeElem(0, e)
fmt.Fprint(b.w, "\n")
}
}
}

602
vendor/golang.org/x/text/unicode/cldr/resolve.go generated vendored
View File

@ -1,602 +0,0 @@
// 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 cldr
// This file implements the various inheritance constructs defined by LDML.
// See https://www.unicode.org/reports/tr35/#Inheritance_and_Validity
// for more details.
import (
"fmt"
"log"
"reflect"
"regexp"
"sort"
"strings"
)
// fieldIter iterates over fields in a struct. It includes
// fields of embedded structs.
type fieldIter struct {
v reflect.Value
index, n []int
}
func iter(v reflect.Value) fieldIter {
if v.Kind() != reflect.Struct {
log.Panicf("value %v must be a struct", v)
}
i := fieldIter{
v: v,
index: []int{0},
n: []int{v.NumField()},
}
i.descent()
return i
}
func (i *fieldIter) descent() {
for f := i.field(); f.Anonymous && f.Type.NumField() > 0; f = i.field() {
i.index = append(i.index, 0)
i.n = append(i.n, f.Type.NumField())
}
}
func (i *fieldIter) done() bool {
return len(i.index) == 1 && i.index[0] >= i.n[0]
}
func skip(f reflect.StructField) bool {
return !f.Anonymous && (f.Name[0] < 'A' || f.Name[0] > 'Z')
}
func (i *fieldIter) next() {
for {
k := len(i.index) - 1
i.index[k]++
if i.index[k] < i.n[k] {
if !skip(i.field()) {
break
}
} else {
if k == 0 {
return
}
i.index = i.index[:k]
i.n = i.n[:k]
}
}
i.descent()
}
func (i *fieldIter) value() reflect.Value {
return i.v.FieldByIndex(i.index)
}
func (i *fieldIter) field() reflect.StructField {
return i.v.Type().FieldByIndex(i.index)
}
type visitor func(v reflect.Value) error
var stopDescent = fmt.Errorf("do not recurse")
func (f visitor) visit(x interface{}) error {
return f.visitRec(reflect.ValueOf(x))
}
// visit recursively calls f on all nodes in v.
func (f visitor) visitRec(v reflect.Value) error {
if v.Kind() == reflect.Ptr {
if v.IsNil() {
return nil
}
return f.visitRec(v.Elem())
}
if err := f(v); err != nil {
if err == stopDescent {
return nil
}
return err
}
switch v.Kind() {
case reflect.Struct:
for i := iter(v); !i.done(); i.next() {
if err := f.visitRec(i.value()); err != nil {
return err
}
}
case reflect.Slice:
for i := 0; i < v.Len(); i++ {
if err := f.visitRec(v.Index(i)); err != nil {
return err
}
}
}
return nil
}
// getPath is used for error reporting purposes only.
func getPath(e Elem) string {
if e == nil {
return "<nil>"
}
if e.enclosing() == nil {
return e.GetCommon().name
}
if e.GetCommon().Type == "" {
return fmt.Sprintf("%s.%s", getPath(e.enclosing()), e.GetCommon().name)
}
return fmt.Sprintf("%s.%s[type=%s]", getPath(e.enclosing()), e.GetCommon().name, e.GetCommon().Type)
}
// xmlName returns the xml name of the element or attribute
func xmlName(f reflect.StructField) (name string, attr bool) {
tags := strings.Split(f.Tag.Get("xml"), ",")
for _, s := range tags {
attr = attr || s == "attr"
}
return tags[0], attr
}
func findField(v reflect.Value, key string) (reflect.Value, error) {
v = reflect.Indirect(v)
for i := iter(v); !i.done(); i.next() {
if n, _ := xmlName(i.field()); n == key {
return i.value(), nil
}
}
return reflect.Value{}, fmt.Errorf("cldr: no field %q in element %#v", key, v.Interface())
}
var xpathPart = regexp.MustCompile(`(\pL+)(?:\[@(\pL+)='([\w-]+)'\])?`)
func walkXPath(e Elem, path string) (res Elem, err error) {
for _, c := range strings.Split(path, "/") {
if c == ".." {
if e = e.enclosing(); e == nil {
panic("path ..")
return nil, fmt.Errorf(`cldr: ".." moves past root in path %q`, path)
}
continue
} else if c == "" {
continue
}
m := xpathPart.FindStringSubmatch(c)
if len(m) == 0 || len(m[0]) != len(c) {
return nil, fmt.Errorf("cldr: syntax error in path component %q", c)
}
v, err := findField(reflect.ValueOf(e), m[1])
if err != nil {
return nil, err
}
switch v.Kind() {
case reflect.Slice:
i := 0
if m[2] != "" || v.Len() > 1 {
if m[2] == "" {
m[2] = "type"
if m[3] = e.GetCommon().Default(); m[3] == "" {
return nil, fmt.Errorf("cldr: type selector or default value needed for element %s", m[1])
}
}
for ; i < v.Len(); i++ {
vi := v.Index(i)
key, err := findField(vi.Elem(), m[2])
if err != nil {
return nil, err
}
key = reflect.Indirect(key)
if key.Kind() == reflect.String && key.String() == m[3] {
break
}
}
}
if i == v.Len() || v.Index(i).IsNil() {
return nil, fmt.Errorf("no %s found with %s==%s", m[1], m[2], m[3])
}
e = v.Index(i).Interface().(Elem)
case reflect.Ptr:
if v.IsNil() {
return nil, fmt.Errorf("cldr: element %q not found within element %q", m[1], e.GetCommon().name)
}
var ok bool
if e, ok = v.Interface().(Elem); !ok {
return nil, fmt.Errorf("cldr: %q is not an XML element", m[1])
} else if m[2] != "" || m[3] != "" {
return nil, fmt.Errorf("cldr: no type selector allowed for element %s", m[1])
}
default:
return nil, fmt.Errorf("cldr: %q is not an XML element", m[1])
}
}
return e, nil
}
const absPrefix = "//ldml/"
func (cldr *CLDR) resolveAlias(e Elem, src, path string) (res Elem, err error) {
if src != "locale" {
if !strings.HasPrefix(path, absPrefix) {
return nil, fmt.Errorf("cldr: expected absolute path, found %q", path)
}
path = path[len(absPrefix):]
if e, err = cldr.resolve(src); err != nil {
return nil, err
}
}
return walkXPath(e, path)
}
func (cldr *CLDR) resolveAndMergeAlias(e Elem) error {
alias := e.GetCommon().Alias
if alias == nil {
return nil
}
a, err := cldr.resolveAlias(e, alias.Source, alias.Path)
if err != nil {
return fmt.Errorf("%v: error evaluating path %q: %v", getPath(e), alias.Path, err)
}
// Ensure alias node was already evaluated. TODO: avoid double evaluation.
err = cldr.resolveAndMergeAlias(a)
v := reflect.ValueOf(e).Elem()
for i := iter(reflect.ValueOf(a).Elem()); !i.done(); i.next() {
if vv := i.value(); vv.Kind() != reflect.Ptr || !vv.IsNil() {
if _, attr := xmlName(i.field()); !attr {
v.FieldByIndex(i.index).Set(vv)
}
}
}
return err
}
func (cldr *CLDR) aliasResolver() visitor {
return func(v reflect.Value) (err error) {
if e, ok := v.Addr().Interface().(Elem); ok {
err = cldr.resolveAndMergeAlias(e)
if err == nil && blocking[e.GetCommon().name] {
return stopDescent
}
}
return err
}
}
// elements within blocking elements do not inherit.
// Taken from CLDR's supplementalMetaData.xml.
var blocking = map[string]bool{
"identity": true,
"supplementalData": true,
"cldrTest": true,
"collation": true,
"transform": true,
}
// Distinguishing attributes affect inheritance; two elements with different
// distinguishing attributes are treated as different for purposes of inheritance,
// except when such attributes occur in the indicated elements.
// Taken from CLDR's supplementalMetaData.xml.
var distinguishing = map[string][]string{
"key": nil,
"request_id": nil,
"id": nil,
"registry": nil,
"alt": nil,
"iso4217": nil,
"iso3166": nil,
"mzone": nil,
"from": nil,
"to": nil,
"type": []string{
"abbreviationFallback",
"default",
"mapping",
"measurementSystem",
"preferenceOrdering",
},
"numberSystem": nil,
}
func in(set []string, s string) bool {
for _, v := range set {
if v == s {
return true
}
}
return false
}
// attrKey computes a key based on the distinguishable attributes of
// an element and its values.
func attrKey(v reflect.Value, exclude ...string) string {
parts := []string{}
ename := v.Interface().(Elem).GetCommon().name
v = v.Elem()
for i := iter(v); !i.done(); i.next() {
if name, attr := xmlName(i.field()); attr {
if except, ok := distinguishing[name]; ok && !in(exclude, name) && !in(except, ename) {
v := i.value()
if v.Kind() == reflect.Ptr {
v = v.Elem()
}
if v.IsValid() {
parts = append(parts, fmt.Sprintf("%s=%s", name, v.String()))
}
}
}
}
sort.Strings(parts)
return strings.Join(parts, ";")
}
// Key returns a key for e derived from all distinguishing attributes
// except those specified by exclude.
func Key(e Elem, exclude ...string) string {
return attrKey(reflect.ValueOf(e), exclude...)
}
// linkEnclosing sets the enclosing element as well as the name
// for all sub-elements of child, recursively.
func linkEnclosing(parent, child Elem) {
child.setEnclosing(parent)
v := reflect.ValueOf(child).Elem()
for i := iter(v); !i.done(); i.next() {
vf := i.value()
if vf.Kind() == reflect.Slice {
for j := 0; j < vf.Len(); j++ {
linkEnclosing(child, vf.Index(j).Interface().(Elem))
}
} else if vf.Kind() == reflect.Ptr && !vf.IsNil() && vf.Elem().Kind() == reflect.Struct {
linkEnclosing(child, vf.Interface().(Elem))
}
}
}
func setNames(e Elem, name string) {
e.setName(name)
v := reflect.ValueOf(e).Elem()
for i := iter(v); !i.done(); i.next() {
vf := i.value()
name, _ = xmlName(i.field())
if vf.Kind() == reflect.Slice {
for j := 0; j < vf.Len(); j++ {
setNames(vf.Index(j).Interface().(Elem), name)
}
} else if vf.Kind() == reflect.Ptr && !vf.IsNil() && vf.Elem().Kind() == reflect.Struct {
setNames(vf.Interface().(Elem), name)
}
}
}
// deepCopy copies elements of v recursively. All elements of v that may
// be modified by inheritance are explicitly copied.
func deepCopy(v reflect.Value) reflect.Value {
switch v.Kind() {
case reflect.Ptr:
if v.IsNil() || v.Elem().Kind() != reflect.Struct {
return v
}
nv := reflect.New(v.Elem().Type())
nv.Elem().Set(v.Elem())
deepCopyRec(nv.Elem(), v.Elem())
return nv
case reflect.Slice:
nv := reflect.MakeSlice(v.Type(), v.Len(), v.Len())
for i := 0; i < v.Len(); i++ {
deepCopyRec(nv.Index(i), v.Index(i))
}
return nv
}
panic("deepCopy: must be called with pointer or slice")
}
// deepCopyRec is only called by deepCopy.
func deepCopyRec(nv, v reflect.Value) {
if v.Kind() == reflect.Struct {
t := v.Type()
for i := 0; i < v.NumField(); i++ {
if name, attr := xmlName(t.Field(i)); name != "" && !attr {
deepCopyRec(nv.Field(i), v.Field(i))
}
}
} else {
nv.Set(deepCopy(v))
}
}
// newNode is used to insert a missing node during inheritance.
func (cldr *CLDR) newNode(v, enc reflect.Value) reflect.Value {
n := reflect.New(v.Type())
for i := iter(v); !i.done(); i.next() {
if name, attr := xmlName(i.field()); name == "" || attr {
n.Elem().FieldByIndex(i.index).Set(i.value())
}
}
n.Interface().(Elem).GetCommon().setEnclosing(enc.Addr().Interface().(Elem))
return n
}
// v, parent must be pointers to struct
func (cldr *CLDR) inheritFields(v, parent reflect.Value) (res reflect.Value, err error) {
t := v.Type()
nv := reflect.New(t)
nv.Elem().Set(v)
for i := iter(v); !i.done(); i.next() {
vf := i.value()
f := i.field()
name, attr := xmlName(f)
if name == "" || attr {
continue
}
pf := parent.FieldByIndex(i.index)
if blocking[name] {
if vf.IsNil() {
vf = pf
}
nv.Elem().FieldByIndex(i.index).Set(deepCopy(vf))
continue
}
switch f.Type.Kind() {
case reflect.Ptr:
if f.Type.Elem().Kind() == reflect.Struct {
if !vf.IsNil() {
if vf, err = cldr.inheritStructPtr(vf, pf); err != nil {
return reflect.Value{}, err
}
vf.Interface().(Elem).setEnclosing(nv.Interface().(Elem))
nv.Elem().FieldByIndex(i.index).Set(vf)
} else if !pf.IsNil() {
n := cldr.newNode(pf.Elem(), v)
if vf, err = cldr.inheritStructPtr(n, pf); err != nil {
return reflect.Value{}, err
}
vf.Interface().(Elem).setEnclosing(nv.Interface().(Elem))
nv.Elem().FieldByIndex(i.index).Set(vf)
}
}
case reflect.Slice:
vf, err := cldr.inheritSlice(nv.Elem(), vf, pf)
if err != nil {
return reflect.Zero(t), err
}
nv.Elem().FieldByIndex(i.index).Set(vf)
}
}
return nv, nil
}
func root(e Elem) *LDML {
for ; e.enclosing() != nil; e = e.enclosing() {
}
return e.(*LDML)
}
// inheritStructPtr first merges possible aliases in with v and then inherits
// any underspecified elements from parent.
func (cldr *CLDR) inheritStructPtr(v, parent reflect.Value) (r reflect.Value, err error) {
if !v.IsNil() {
e := v.Interface().(Elem).GetCommon()
alias := e.Alias
if alias == nil && !parent.IsNil() {
alias = parent.Interface().(Elem).GetCommon().Alias
}
if alias != nil {
a, err := cldr.resolveAlias(v.Interface().(Elem), alias.Source, alias.Path)
if a != nil {
if v, err = cldr.inheritFields(v.Elem(), reflect.ValueOf(a).Elem()); err != nil {
return reflect.Value{}, err
}
}
}
if !parent.IsNil() {
return cldr.inheritFields(v.Elem(), parent.Elem())
}
} else if parent.IsNil() {
panic("should not reach here")
}
return v, nil
}
// Must be slice of struct pointers.
func (cldr *CLDR) inheritSlice(enc, v, parent reflect.Value) (res reflect.Value, err error) {
t := v.Type()
index := make(map[string]reflect.Value)
if !v.IsNil() {
for i := 0; i < v.Len(); i++ {
vi := v.Index(i)
key := attrKey(vi)
index[key] = vi
}
}
if !parent.IsNil() {
for i := 0; i < parent.Len(); i++ {
vi := parent.Index(i)
key := attrKey(vi)
if w, ok := index[key]; ok {
index[key], err = cldr.inheritStructPtr(w, vi)
} else {
n := cldr.newNode(vi.Elem(), enc)
index[key], err = cldr.inheritStructPtr(n, vi)
}
index[key].Interface().(Elem).setEnclosing(enc.Addr().Interface().(Elem))
if err != nil {
return v, err
}
}
}
keys := make([]string, 0, len(index))
for k, _ := range index {
keys = append(keys, k)
}
sort.Strings(keys)
sl := reflect.MakeSlice(t, len(index), len(index))
for i, k := range keys {
sl.Index(i).Set(index[k])
}
return sl, nil
}
func parentLocale(loc string) string {
parts := strings.Split(loc, "_")
if len(parts) == 1 {
return "root"
}
parts = parts[:len(parts)-1]
key := strings.Join(parts, "_")
return key
}
func (cldr *CLDR) resolve(loc string) (res *LDML, err error) {
if r := cldr.resolved[loc]; r != nil {
return r, nil
}
x := cldr.RawLDML(loc)
if x == nil {
return nil, fmt.Errorf("cldr: unknown locale %q", loc)
}
var v reflect.Value
if loc == "root" {
x = deepCopy(reflect.ValueOf(x)).Interface().(*LDML)
linkEnclosing(nil, x)
err = cldr.aliasResolver().visit(x)
} else {
key := parentLocale(loc)
var parent *LDML
for ; cldr.locale[key] == nil; key = parentLocale(key) {
}
if parent, err = cldr.resolve(key); err != nil {
return nil, err
}
v, err = cldr.inheritFields(reflect.ValueOf(x).Elem(), reflect.ValueOf(parent).Elem())
x = v.Interface().(*LDML)
linkEnclosing(nil, x)
}
if err != nil {
return nil, err
}
cldr.resolved[loc] = x
return x, err
}
// finalize finalizes the initialization of the raw LDML structs. It also
// removed unwanted fields, as specified by filter, so that they will not
// be unnecessarily evaluated.
func (cldr *CLDR) finalize(filter []string) {
for _, x := range cldr.locale {
if filter != nil {
v := reflect.ValueOf(x).Elem()
t := v.Type()
for i := 0; i < v.NumField(); i++ {
f := t.Field(i)
name, _ := xmlName(f)
if name != "" && name != "identity" && !in(filter, name) {
v.Field(i).Set(reflect.Zero(f.Type))
}
}
}
linkEnclosing(nil, x) // for resolving aliases and paths
setNames(x, "ldml")
}
}

144
vendor/golang.org/x/text/unicode/cldr/slice.go generated vendored
View File

@ -1,144 +0,0 @@
// 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 cldr
import (
"fmt"
"reflect"
"sort"
)
// Slice provides utilities for modifying slices of elements.
// It can be wrapped around any slice of which the element type implements
// interface Elem.
type Slice struct {
ptr reflect.Value
typ reflect.Type
}
// Value returns the reflect.Value of the underlying slice.
func (s *Slice) Value() reflect.Value {
return s.ptr.Elem()
}
// MakeSlice wraps a pointer to a slice of Elems.
// It replaces the array pointed to by the slice so that subsequent modifications
// do not alter the data in a CLDR type.
// It panics if an incorrect type is passed.
func MakeSlice(slicePtr interface{}) Slice {
ptr := reflect.ValueOf(slicePtr)
if ptr.Kind() != reflect.Ptr {
panic(fmt.Sprintf("MakeSlice: argument must be pointer to slice, found %v", ptr.Type()))
}
sl := ptr.Elem()
if sl.Kind() != reflect.Slice {
panic(fmt.Sprintf("MakeSlice: argument must point to a slice, found %v", sl.Type()))
}
intf := reflect.TypeOf((*Elem)(nil)).Elem()
if !sl.Type().Elem().Implements(intf) {
panic(fmt.Sprintf("MakeSlice: element type of slice (%v) does not implement Elem", sl.Type().Elem()))
}
nsl := reflect.MakeSlice(sl.Type(), sl.Len(), sl.Len())
reflect.Copy(nsl, sl)
sl.Set(nsl)
return Slice{
ptr: ptr,
typ: sl.Type().Elem().Elem(),
}
}
func (s Slice) indexForAttr(a string) []int {
for i := iter(reflect.Zero(s.typ)); !i.done(); i.next() {
if n, _ := xmlName(i.field()); n == a {
return i.index
}
}
panic(fmt.Sprintf("MakeSlice: no attribute %q for type %v", a, s.typ))
}
// Filter filters s to only include elements for which fn returns true.
func (s Slice) Filter(fn func(e Elem) bool) {
k := 0
sl := s.Value()
for i := 0; i < sl.Len(); i++ {
vi := sl.Index(i)
if fn(vi.Interface().(Elem)) {
sl.Index(k).Set(vi)
k++
}
}
sl.Set(sl.Slice(0, k))
}
// Group finds elements in s for which fn returns the same value and groups
// them in a new Slice.
func (s Slice) Group(fn func(e Elem) string) []Slice {
m := make(map[string][]reflect.Value)
sl := s.Value()
for i := 0; i < sl.Len(); i++ {
vi := sl.Index(i)
key := fn(vi.Interface().(Elem))
m[key] = append(m[key], vi)
}
keys := []string{}
for k, _ := range m {
keys = append(keys, k)
}
sort.Strings(keys)
res := []Slice{}
for _, k := range keys {
nsl := reflect.New(sl.Type())
nsl.Elem().Set(reflect.Append(nsl.Elem(), m[k]...))
res = append(res, MakeSlice(nsl.Interface()))
}
return res
}
// SelectAnyOf filters s to contain only elements for which attr matches
// any of the values.
func (s Slice) SelectAnyOf(attr string, values ...string) {
index := s.indexForAttr(attr)
s.Filter(func(e Elem) bool {
vf := reflect.ValueOf(e).Elem().FieldByIndex(index)
return in(values, vf.String())
})
}
// SelectOnePerGroup filters s to include at most one element e per group of
// elements matching Key(attr), where e has an attribute a that matches any
// the values in v.
// If more than one element in a group matches a value in v preference
// is given to the element that matches the first value in v.
func (s Slice) SelectOnePerGroup(a string, v []string) {
index := s.indexForAttr(a)
grouped := s.Group(func(e Elem) string { return Key(e, a) })
sl := s.Value()
sl.Set(sl.Slice(0, 0))
for _, g := range grouped {
e := reflect.Value{}
found := len(v)
gsl := g.Value()
for i := 0; i < gsl.Len(); i++ {
vi := gsl.Index(i).Elem().FieldByIndex(index)
j := 0
for ; j < len(v) && v[j] != vi.String(); j++ {
}
if j < found {
found = j
e = gsl.Index(i)
}
}
if found < len(v) {
sl.Set(reflect.Append(sl, e))
}
}
}
// SelectDraft drops all elements from the list with a draft level smaller than d
// and selects the highest draft level of the remaining.
// This method assumes that the input CLDR is canonicalized.
func (s Slice) SelectDraft(d Draft) {
s.SelectOnePerGroup("draft", drafts[len(drafts)-2-int(d):])
}

1494
vendor/golang.org/x/text/unicode/cldr/xml.go generated vendored
View File

File diff suppressed because it is too large Load Diff

986
vendor/golang.org/x/text/unicode/norm/maketables.go generated vendored
View File

@ -1,986 +0,0 @@
// 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.
// +build ignore
// Normalization table generator.
// Data read from the web.
// See forminfo.go for a description of the trie values associated with each rune.
package main
import (
"bytes"
"encoding/binary"
"flag"
"fmt"
"io"
"log"
"sort"
"strconv"
"strings"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/triegen"
"golang.org/x/text/internal/ucd"
)
func main() {
gen.Init()
loadUnicodeData()
compactCCC()
loadCompositionExclusions()
completeCharFields(FCanonical)
completeCharFields(FCompatibility)
computeNonStarterCounts()
verifyComputed()
printChars()
testDerived()
printTestdata()
makeTables()
}
var (
tablelist = flag.String("tables",
"all",
"comma-separated list of which tables to generate; "+
"can be 'decomp', 'recomp', 'info' and 'all'")
test = flag.Bool("test",
false,
"test existing tables against DerivedNormalizationProps and generate test data for regression testing")
verbose = flag.Bool("verbose",
false,
"write data to stdout as it is parsed")
)
const MaxChar = 0x10FFFF // anything above this shouldn't exist
// Quick Check properties of runes allow us to quickly
// determine whether a rune may occur in a normal form.
// For a given normal form, a rune may be guaranteed to occur
// verbatim (QC=Yes), may or may not combine with another
// rune (QC=Maybe), or may not occur (QC=No).
type QCResult int
const (
QCUnknown QCResult = iota
QCYes
QCNo
QCMaybe
)
func (r QCResult) String() string {
switch r {
case QCYes:
return "Yes"
case QCNo:
return "No"
case QCMaybe:
return "Maybe"
}
return "***UNKNOWN***"
}
const (
FCanonical = iota // NFC or NFD
FCompatibility // NFKC or NFKD
FNumberOfFormTypes
)
const (
MComposed = iota // NFC or NFKC
MDecomposed // NFD or NFKD
MNumberOfModes
)
// This contains only the properties we're interested in.
type Char struct {
name string
codePoint rune // if zero, this index is not a valid code point.
ccc uint8 // canonical combining class
origCCC uint8
excludeInComp bool // from CompositionExclusions.txt
compatDecomp bool // it has a compatibility expansion
nTrailingNonStarters uint8
nLeadingNonStarters uint8 // must be equal to trailing if non-zero
forms [FNumberOfFormTypes]FormInfo // For FCanonical and FCompatibility
state State
}
var chars = make([]Char, MaxChar+1)
var cccMap = make(map[uint8]uint8)
func (c Char) String() string {
buf := new(bytes.Buffer)
fmt.Fprintf(buf, "%U [%s]:\n", c.codePoint, c.name)
fmt.Fprintf(buf, " ccc: %v\n", c.ccc)
fmt.Fprintf(buf, " excludeInComp: %v\n", c.excludeInComp)
fmt.Fprintf(buf, " compatDecomp: %v\n", c.compatDecomp)
fmt.Fprintf(buf, " state: %v\n", c.state)
fmt.Fprintf(buf, " NFC:\n")
fmt.Fprint(buf, c.forms[FCanonical])
fmt.Fprintf(buf, " NFKC:\n")
fmt.Fprint(buf, c.forms[FCompatibility])
return buf.String()
}
// 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;;;;;
// parseCharacter keeps a state variable indicating the weirdness.
type State int
const (
SNormal State = iota // known to be zero for the type
SFirst
SLast
SMissing
)
var lastChar = rune('\u0000')
func (c Char) isValid() bool {
return c.codePoint != 0 && c.state != SMissing
}
type FormInfo struct {
quickCheck [MNumberOfModes]QCResult // index: MComposed or MDecomposed
verified [MNumberOfModes]bool // index: MComposed or MDecomposed
combinesForward bool // May combine with rune on the right
combinesBackward bool // May combine with rune on the left
isOneWay bool // Never appears in result
inDecomp bool // Some decompositions result in this char.
decomp Decomposition
expandedDecomp Decomposition
}
func (f FormInfo) String() string {
buf := bytes.NewBuffer(make([]byte, 0))
fmt.Fprintf(buf, " quickCheck[C]: %v\n", f.quickCheck[MComposed])
fmt.Fprintf(buf, " quickCheck[D]: %v\n", f.quickCheck[MDecomposed])
fmt.Fprintf(buf, " cmbForward: %v\n", f.combinesForward)
fmt.Fprintf(buf, " cmbBackward: %v\n", f.combinesBackward)
fmt.Fprintf(buf, " isOneWay: %v\n", f.isOneWay)
fmt.Fprintf(buf, " inDecomp: %v\n", f.inDecomp)
fmt.Fprintf(buf, " decomposition: %X\n", f.decomp)
fmt.Fprintf(buf, " expandedDecomp: %X\n", f.expandedDecomp)
return buf.String()
}
type Decomposition []rune
func parseDecomposition(s string, skipfirst bool) (a []rune, err error) {
decomp := strings.Split(s, " ")
if len(decomp) > 0 && skipfirst {
decomp = decomp[1:]
}
for _, d := range decomp {
point, err := strconv.ParseUint(d, 16, 64)
if err != nil {
return a, err
}
a = append(a, rune(point))
}
return a, nil
}
func loadUnicodeData() {
f := gen.OpenUCDFile("UnicodeData.txt")
defer f.Close()
p := ucd.New(f)
for p.Next() {
r := p.Rune(ucd.CodePoint)
char := &chars[r]
char.ccc = uint8(p.Uint(ucd.CanonicalCombiningClass))
decmap := p.String(ucd.DecompMapping)
exp, err := parseDecomposition(decmap, false)
isCompat := false
if err != nil {
if len(decmap) > 0 {
exp, err = parseDecomposition(decmap, true)
if err != nil {
log.Fatalf(`%U: bad decomp |%v|: "%s"`, r, decmap, err)
}
isCompat = true
}
}
char.name = p.String(ucd.Name)
char.codePoint = r
char.forms[FCompatibility].decomp = exp
if !isCompat {
char.forms[FCanonical].decomp = exp
} else {
char.compatDecomp = true
}
if len(decmap) > 0 {
char.forms[FCompatibility].decomp = exp
}
}
if err := p.Err(); err != nil {
log.Fatal(err)
}
}
// compactCCC converts the sparse set of CCC values to a continguous one,
// reducing the number of bits needed from 8 to 6.
func compactCCC() {
m := make(map[uint8]uint8)
for i := range chars {
c := &chars[i]
m[c.ccc] = 0
}
cccs := []int{}
for v, _ := range m {
cccs = append(cccs, int(v))
}
sort.Ints(cccs)
for i, c := range cccs {
cccMap[uint8(i)] = uint8(c)
m[uint8(c)] = uint8(i)
}
for i := range chars {
c := &chars[i]
c.origCCC = c.ccc
c.ccc = m[c.ccc]
}
if len(m) >= 1<<6 {
log.Fatalf("too many difference CCC values: %d >= 64", len(m))
}
}
// CompositionExclusions.txt has form:
// 0958 # ...
// See https://unicode.org/reports/tr44/ for full explanation
func loadCompositionExclusions() {
f := gen.OpenUCDFile("CompositionExclusions.txt")
defer f.Close()
p := ucd.New(f)
for p.Next() {
c := &chars[p.Rune(0)]
if c.excludeInComp {
log.Fatalf("%U: Duplicate entry in exclusions.", c.codePoint)
}
c.excludeInComp = true
}
if e := p.Err(); e != nil {
log.Fatal(e)
}
}
// hasCompatDecomp returns true if any of the recursive
// decompositions contains a compatibility expansion.
// In this case, the character may not occur in NFK*.
func hasCompatDecomp(r rune) bool {
c := &chars[r]
if c.compatDecomp {
return true
}
for _, d := range c.forms[FCompatibility].decomp {
if hasCompatDecomp(d) {
return true
}
}
return false
}
// Hangul related constants.
const (
HangulBase = 0xAC00
HangulEnd = 0xD7A4 // hangulBase + Jamo combinations (19 * 21 * 28)
JamoLBase = 0x1100
JamoLEnd = 0x1113
JamoVBase = 0x1161
JamoVEnd = 0x1176
JamoTBase = 0x11A8
JamoTEnd = 0x11C3
JamoLVTCount = 19 * 21 * 28
JamoTCount = 28
)
func isHangul(r rune) bool {
return HangulBase <= r && r < HangulEnd
}
func isHangulWithoutJamoT(r rune) bool {
if !isHangul(r) {
return false
}
r -= HangulBase
return r < JamoLVTCount && r%JamoTCount == 0
}
func ccc(r rune) uint8 {
return chars[r].ccc
}
// Insert a rune in a buffer, ordered by Canonical Combining Class.
func insertOrdered(b Decomposition, r rune) Decomposition {
n := len(b)
b = append(b, 0)
cc := ccc(r)
if cc > 0 {
// Use bubble sort.
for ; n > 0; n-- {
if ccc(b[n-1]) <= cc {
break
}
b[n] = b[n-1]
}
}
b[n] = r
return b
}
// Recursively decompose.
func decomposeRecursive(form int, r rune, d Decomposition) Decomposition {
dcomp := chars[r].forms[form].decomp
if len(dcomp) == 0 {
return insertOrdered(d, r)
}
for _, c := range dcomp {
d = decomposeRecursive(form, c, d)
}
return d
}
func completeCharFields(form int) {
// Phase 0: pre-expand decomposition.
for i := range chars {
f := &chars[i].forms[form]
if len(f.decomp) == 0 {
continue
}
exp := make(Decomposition, 0)
for _, c := range f.decomp {
exp = decomposeRecursive(form, c, exp)
}
f.expandedDecomp = exp
}
// Phase 1: composition exclusion, mark decomposition.
for i := range chars {
c := &chars[i]
f := &c.forms[form]
// Marks script-specific exclusions and version restricted.
f.isOneWay = c.excludeInComp
// Singletons
f.isOneWay = f.isOneWay || len(f.decomp) == 1
// Non-starter decompositions
if len(f.decomp) > 1 {
chk := c.ccc != 0 || chars[f.decomp[0]].ccc != 0
f.isOneWay = f.isOneWay || chk
}
// Runes that decompose into more than two runes.
f.isOneWay = f.isOneWay || len(f.decomp) > 2
if form == FCompatibility {
f.isOneWay = f.isOneWay || hasCompatDecomp(c.codePoint)
}
for _, r := range f.decomp {
chars[r].forms[form].inDecomp = true
}
}
// Phase 2: forward and backward combining.
for i := range chars {
c := &chars[i]
f := &c.forms[form]
if !f.isOneWay && len(f.decomp) == 2 {
f0 := &chars[f.decomp[0]].forms[form]
f1 := &chars[f.decomp[1]].forms[form]
if !f0.isOneWay {
f0.combinesForward = true
}
if !f1.isOneWay {
f1.combinesBackward = true
}
}
if isHangulWithoutJamoT(rune(i)) {
f.combinesForward = true
}
}
// Phase 3: quick check values.
for i := range chars {
c := &chars[i]
f := &c.forms[form]
switch {
case len(f.decomp) > 0:
f.quickCheck[MDecomposed] = QCNo
case isHangul(rune(i)):
f.quickCheck[MDecomposed] = QCNo
default:
f.quickCheck[MDecomposed] = QCYes
}
switch {
case f.isOneWay:
f.quickCheck[MComposed] = QCNo
case (i & 0xffff00) == JamoLBase:
f.quickCheck[MComposed] = QCYes
if JamoLBase <= i && i < JamoLEnd {
f.combinesForward = true
}
if JamoVBase <= i && i < JamoVEnd {
f.quickCheck[MComposed] = QCMaybe
f.combinesBackward = true
f.combinesForward = true
}
if JamoTBase <= i && i < JamoTEnd {
f.quickCheck[MComposed] = QCMaybe
f.combinesBackward = true
}
case !f.combinesBackward:
f.quickCheck[MComposed] = QCYes
default:
f.quickCheck[MComposed] = QCMaybe
}
}
}
func computeNonStarterCounts() {
// Phase 4: leading and trailing non-starter count
for i := range chars {
c := &chars[i]
runes := []rune{rune(i)}
// We always use FCompatibility so that the CGJ insertion points do not
// change for repeated normalizations with different forms.
if exp := c.forms[FCompatibility].expandedDecomp; len(exp) > 0 {
runes = exp
}
// We consider runes that combine backwards to be non-starters for the
// purpose of Stream-Safe Text Processing.
for _, r := range runes {
if cr := &chars[r]; cr.ccc == 0 && !cr.forms[FCompatibility].combinesBackward {
break
}
c.nLeadingNonStarters++
}
for i := len(runes) - 1; i >= 0; i-- {
if cr := &chars[runes[i]]; cr.ccc == 0 && !cr.forms[FCompatibility].combinesBackward {
break
}
c.nTrailingNonStarters++
}
if c.nTrailingNonStarters > 3 {
log.Fatalf("%U: Decomposition with more than 3 (%d) trailing modifiers (%U)", i, c.nTrailingNonStarters, runes)
}
if isHangul(rune(i)) {
c.nTrailingNonStarters = 2
if isHangulWithoutJamoT(rune(i)) {
c.nTrailingNonStarters = 1
}
}
if l, t := c.nLeadingNonStarters, c.nTrailingNonStarters; l > 0 && l != t {
log.Fatalf("%U: number of leading and trailing non-starters should be equal (%d vs %d)", i, l, t)
}
if t := c.nTrailingNonStarters; t > 3 {
log.Fatalf("%U: number of trailing non-starters is %d > 3", t)
}
}
}
func printBytes(w io.Writer, b []byte, name string) {
fmt.Fprintf(w, "// %s: %d bytes\n", name, len(b))
fmt.Fprintf(w, "var %s = [...]byte {", name)
for i, c := range b {
switch {
case i%64 == 0:
fmt.Fprintf(w, "\n// Bytes %x - %x\n", i, i+63)
case i%8 == 0:
fmt.Fprintf(w, "\n")
}
fmt.Fprintf(w, "0x%.2X, ", c)
}
fmt.Fprint(w, "\n}\n\n")
}
// See forminfo.go for format.
func makeEntry(f *FormInfo, c *Char) uint16 {
e := uint16(0)
if r := c.codePoint; HangulBase <= r && r < HangulEnd {
e |= 0x40
}
if f.combinesForward {
e |= 0x20
}
if f.quickCheck[MDecomposed] == QCNo {
e |= 0x4
}
switch f.quickCheck[MComposed] {
case QCYes:
case QCNo:
e |= 0x10
case QCMaybe:
e |= 0x18
default:
log.Fatalf("Illegal quickcheck value %v.", f.quickCheck[MComposed])
}
e |= uint16(c.nTrailingNonStarters)
return e
}
// decompSet keeps track of unique decompositions, grouped by whether
// the decomposition is followed by a trailing and/or leading CCC.
type decompSet [7]map[string]bool
const (
normalDecomp = iota
firstMulti
firstCCC
endMulti
firstLeadingCCC
firstCCCZeroExcept
firstStarterWithNLead
lastDecomp
)
var cname = []string{"firstMulti", "firstCCC", "endMulti", "firstLeadingCCC", "firstCCCZeroExcept", "firstStarterWithNLead", "lastDecomp"}
func makeDecompSet() decompSet {
m := decompSet{}
for i := range m {
m[i] = make(map[string]bool)
}
return m
}
func (m *decompSet) insert(key int, s string) {
m[key][s] = true
}
func printCharInfoTables(w io.Writer) int {
mkstr := func(r rune, f *FormInfo) (int, string) {
d := f.expandedDecomp
s := string([]rune(d))
if max := 1 << 6; len(s) >= max {
const msg = "%U: too many bytes in decomposition: %d >= %d"
log.Fatalf(msg, r, len(s), max)
}
head := uint8(len(s))
if f.quickCheck[MComposed] != QCYes {
head |= 0x40
}
if f.combinesForward {
head |= 0x80
}
s = string([]byte{head}) + s
lccc := ccc(d[0])
tccc := ccc(d[len(d)-1])
cc := ccc(r)
if cc != 0 && lccc == 0 && tccc == 0 {
log.Fatalf("%U: trailing and leading ccc are 0 for non-zero ccc %d", r, cc)
}
if tccc < lccc && lccc != 0 {
const msg = "%U: lccc (%d) must be <= tcc (%d)"
log.Fatalf(msg, r, lccc, tccc)
}
index := normalDecomp
nTrail := chars[r].nTrailingNonStarters
nLead := chars[r].nLeadingNonStarters
if tccc > 0 || lccc > 0 || nTrail > 0 {
tccc <<= 2
tccc |= nTrail
s += string([]byte{tccc})
index = endMulti
for _, r := range d[1:] {
if ccc(r) == 0 {
index = firstCCC
}
}
if lccc > 0 || nLead > 0 {
s += string([]byte{lccc})
if index == firstCCC {
log.Fatalf("%U: multi-segment decomposition not supported for decompositions with leading CCC != 0", r)
}
index = firstLeadingCCC
}
if cc != lccc {
if cc != 0 {
log.Fatalf("%U: for lccc != ccc, expected ccc to be 0; was %d", r, cc)
}
index = firstCCCZeroExcept
}
} else if len(d) > 1 {
index = firstMulti
}
return index, s
}
decompSet := makeDecompSet()
const nLeadStr = "\x00\x01" // 0-byte length and tccc with nTrail.
decompSet.insert(firstStarterWithNLead, nLeadStr)
// Store the uniqued decompositions in a byte buffer,
// preceded by their byte length.
for _, c := range chars {
for _, f := range c.forms {
if len(f.expandedDecomp) == 0 {
continue
}
if f.combinesBackward {
log.Fatalf("%U: combinesBackward and decompose", c.codePoint)
}
index, s := mkstr(c.codePoint, &f)
decompSet.insert(index, s)
}
}
decompositions := bytes.NewBuffer(make([]byte, 0, 10000))
size := 0
positionMap := make(map[string]uint16)
decompositions.WriteString("\000")
fmt.Fprintln(w, "const (")
for i, m := range decompSet {
sa := []string{}
for s := range m {
sa = append(sa, s)
}
sort.Strings(sa)
for _, s := range sa {
p := decompositions.Len()
decompositions.WriteString(s)
positionMap[s] = uint16(p)
}
if cname[i] != "" {
fmt.Fprintf(w, "%s = 0x%X\n", cname[i], decompositions.Len())
}
}
fmt.Fprintln(w, "maxDecomp = 0x8000")
fmt.Fprintln(w, ")")
b := decompositions.Bytes()
printBytes(w, b, "decomps")
size += len(b)
varnames := []string{"nfc", "nfkc"}
for i := 0; i < FNumberOfFormTypes; i++ {
trie := triegen.NewTrie(varnames[i])
for r, c := range chars {
f := c.forms[i]
d := f.expandedDecomp
if len(d) != 0 {
_, key := mkstr(c.codePoint, &f)
trie.Insert(rune(r), uint64(positionMap[key]))
if c.ccc != ccc(d[0]) {
// We assume the lead ccc of a decomposition !=0 in this case.
if ccc(d[0]) == 0 {
log.Fatalf("Expected leading CCC to be non-zero; ccc is %d", c.ccc)
}
}
} else if c.nLeadingNonStarters > 0 && len(f.expandedDecomp) == 0 && c.ccc == 0 && !f.combinesBackward {
// Handle cases where it can't be detected that the nLead should be equal
// to nTrail.
trie.Insert(c.codePoint, uint64(positionMap[nLeadStr]))
} else if v := makeEntry(&f, &c)<<8 | uint16(c.ccc); v != 0 {
trie.Insert(c.codePoint, uint64(0x8000|v))
}
}
sz, err := trie.Gen(w, triegen.Compact(&normCompacter{name: varnames[i]}))
if err != nil {
log.Fatal(err)
}
size += sz
}
return size
}
func contains(sa []string, s string) bool {
for _, a := range sa {
if a == s {
return true
}
}
return false
}
func makeTables() {
w := &bytes.Buffer{}
size := 0
if *tablelist == "" {
return
}
list := strings.Split(*tablelist, ",")
if *tablelist == "all" {
list = []string{"recomp", "info"}
}
// Compute maximum decomposition size.
max := 0
for _, c := range chars {
if n := len(string(c.forms[FCompatibility].expandedDecomp)); n > max {
max = n
}
}
fmt.Fprintln(w, `import "sync"`)
fmt.Fprintln(w)
fmt.Fprintln(w, "const (")
fmt.Fprintln(w, "\t// Version is the Unicode edition from which the tables are derived.")
fmt.Fprintf(w, "\tVersion = %q\n", gen.UnicodeVersion())
fmt.Fprintln(w)
fmt.Fprintln(w, "\t// MaxTransformChunkSize indicates the maximum number of bytes that Transform")
fmt.Fprintln(w, "\t// may need to write atomically for any Form. Making a destination buffer at")
fmt.Fprintln(w, "\t// least this size ensures that Transform can always make progress and that")
fmt.Fprintln(w, "\t// the user does not need to grow the buffer on an ErrShortDst.")
fmt.Fprintf(w, "\tMaxTransformChunkSize = %d+maxNonStarters*4\n", len(string(0x034F))+max)
fmt.Fprintln(w, ")\n")
// Print the CCC remap table.
size += len(cccMap)
fmt.Fprintf(w, "var ccc = [%d]uint8{", len(cccMap))
for i := 0; i < len(cccMap); i++ {
if i%8 == 0 {
fmt.Fprintln(w)
}
fmt.Fprintf(w, "%3d, ", cccMap[uint8(i)])
}
fmt.Fprintln(w, "\n}\n")
if contains(list, "info") {
size += printCharInfoTables(w)
}
if contains(list, "recomp") {
// Note that we use 32 bit keys, instead of 64 bit.
// This clips the bits of three entries, but we know
// this won't cause a collision. The compiler will catch
// any changes made to UnicodeData.txt that introduces
// a collision.
// Note that the recomposition map for NFC and NFKC
// are identical.
// Recomposition map
nrentries := 0
for _, c := range chars {
f := c.forms[FCanonical]
if !f.isOneWay && len(f.decomp) > 0 {
nrentries++
}
}
sz := nrentries * 8
size += sz
fmt.Fprintf(w, "// recompMap: %d bytes (entries only)\n", sz)
fmt.Fprintln(w, "var recompMap map[uint32]rune")
fmt.Fprintln(w, "var recompMapOnce sync.Once\n")
fmt.Fprintln(w, `const recompMapPacked = "" +`)
var buf [8]byte
for i, c := range chars {
f := c.forms[FCanonical]
d := f.decomp
if !f.isOneWay && len(d) > 0 {
key := uint32(uint16(d[0]))<<16 + uint32(uint16(d[1]))
binary.BigEndian.PutUint32(buf[:4], key)
binary.BigEndian.PutUint32(buf[4:], uint32(i))
fmt.Fprintf(w, "\t\t%q + // 0x%.8X: 0x%.8X\n", string(buf[:]), key, uint32(i))
}
}
// hack so we don't have to special case the trailing plus sign
fmt.Fprintf(w, ` ""`)
fmt.Fprintln(w)
}
fmt.Fprintf(w, "// Total size of tables: %dKB (%d bytes)\n", (size+512)/1024, size)
gen.WriteVersionedGoFile("tables.go", "norm", w.Bytes())
}
func printChars() {
if *verbose {
for _, c := range chars {
if !c.isValid() || c.state == SMissing {
continue
}
fmt.Println(c)
}
}
}
// verifyComputed does various consistency tests.
func verifyComputed() {
for i, c := range chars {
for _, f := range c.forms {
isNo := (f.quickCheck[MDecomposed] == QCNo)
if (len(f.decomp) > 0) != isNo && !isHangul(rune(i)) {
log.Fatalf("%U: NF*D QC must be No if rune decomposes", i)
}
isMaybe := f.quickCheck[MComposed] == QCMaybe
if f.combinesBackward != isMaybe {
log.Fatalf("%U: NF*C QC must be Maybe if combinesBackward", i)
}
if len(f.decomp) > 0 && f.combinesForward && isMaybe {
log.Fatalf("%U: NF*C QC must be Yes or No if combinesForward and decomposes", i)
}
if len(f.expandedDecomp) != 0 {
continue
}
if a, b := c.nLeadingNonStarters > 0, (c.ccc > 0 || f.combinesBackward); a != b {
// We accept these runes to be treated differently (it only affects
// segment breaking in iteration, most likely on improper use), but
// reconsider if more characters are added.
// U+FF9E HALFWIDTH KATAKANA VOICED SOUND MARK;Lm;0;L;<narrow> 3099;;;;N;;;;;
// U+FF9F HALFWIDTH KATAKANA SEMI-VOICED SOUND MARK;Lm;0;L;<narrow> 309A;;;;N;;;;;
// U+3133 HANGUL LETTER KIYEOK-SIOS;Lo;0;L;<compat> 11AA;;;;N;HANGUL LETTER GIYEOG SIOS;;;;
// U+318E HANGUL LETTER ARAEAE;Lo;0;L;<compat> 11A1;;;;N;HANGUL LETTER ALAE AE;;;;
// U+FFA3 HALFWIDTH HANGUL LETTER KIYEOK-SIOS;Lo;0;L;<narrow> 3133;;;;N;HALFWIDTH HANGUL LETTER GIYEOG SIOS;;;;
// U+FFDC HALFWIDTH HANGUL LETTER I;Lo;0;L;<narrow> 3163;;;;N;;;;;
if i != 0xFF9E && i != 0xFF9F && !(0x3133 <= i && i <= 0x318E) && !(0xFFA3 <= i && i <= 0xFFDC) {
log.Fatalf("%U: nLead was %v; want %v", i, a, b)
}
}
}
nfc := c.forms[FCanonical]
nfkc := c.forms[FCompatibility]
if nfc.combinesBackward != nfkc.combinesBackward {
log.Fatalf("%U: Cannot combine combinesBackward\n", c.codePoint)
}
}
}
// Use values in DerivedNormalizationProps.txt to compare against the
// values we computed.
// DerivedNormalizationProps.txt has form:
// 00C0..00C5 ; NFD_QC; N # ...
// 0374 ; NFD_QC; N # ...
// See https://unicode.org/reports/tr44/ for full explanation
func testDerived() {
f := gen.OpenUCDFile("DerivedNormalizationProps.txt")
defer f.Close()
p := ucd.New(f)
for p.Next() {
r := p.Rune(0)
c := &chars[r]
var ftype, mode int
qt := p.String(1)
switch qt {
case "NFC_QC":
ftype, mode = FCanonical, MComposed
case "NFD_QC":
ftype, mode = FCanonical, MDecomposed
case "NFKC_QC":
ftype, mode = FCompatibility, MComposed
case "NFKD_QC":
ftype, mode = FCompatibility, MDecomposed
default:
continue
}
var qr QCResult
switch p.String(2) {
case "Y":
qr = QCYes
case "N":
qr = QCNo
case "M":
qr = QCMaybe
default:
log.Fatalf(`Unexpected quick check value "%s"`, p.String(2))
}
if got := c.forms[ftype].quickCheck[mode]; got != qr {
log.Printf("%U: FAILED %s (was %v need %v)\n", r, qt, got, qr)
}
c.forms[ftype].verified[mode] = true
}
if err := p.Err(); err != nil {
log.Fatal(err)
}
// Any unspecified value must be QCYes. Verify this.
for i, c := range chars {
for j, fd := range c.forms {
for k, qr := range fd.quickCheck {
if !fd.verified[k] && qr != QCYes {
m := "%U: FAIL F:%d M:%d (was %v need Yes) %s\n"
log.Printf(m, i, j, k, qr, c.name)
}
}
}
}
}
var testHeader = `const (
Yes = iota
No
Maybe
)
type formData struct {
qc uint8
combinesForward bool
decomposition string
}
type runeData struct {
r rune
ccc uint8
nLead uint8
nTrail uint8
f [2]formData // 0: canonical; 1: compatibility
}
func f(qc uint8, cf bool, dec string) [2]formData {
return [2]formData{{qc, cf, dec}, {qc, cf, dec}}
}
func g(qc, qck uint8, cf, cfk bool, d, dk string) [2]formData {
return [2]formData{{qc, cf, d}, {qck, cfk, dk}}
}
var testData = []runeData{
`
func printTestdata() {
type lastInfo struct {
ccc uint8
nLead uint8
nTrail uint8
f string
}
last := lastInfo{}
w := &bytes.Buffer{}
fmt.Fprintf(w, testHeader)
for r, c := range chars {
f := c.forms[FCanonical]
qc, cf, d := f.quickCheck[MComposed], f.combinesForward, string(f.expandedDecomp)
f = c.forms[FCompatibility]
qck, cfk, dk := f.quickCheck[MComposed], f.combinesForward, string(f.expandedDecomp)
s := ""
if d == dk && qc == qck && cf == cfk {
s = fmt.Sprintf("f(%s, %v, %q)", qc, cf, d)
} else {
s = fmt.Sprintf("g(%s, %s, %v, %v, %q, %q)", qc, qck, cf, cfk, d, dk)
}
current := lastInfo{c.ccc, c.nLeadingNonStarters, c.nTrailingNonStarters, s}
if last != current {
fmt.Fprintf(w, "\t{0x%x, %d, %d, %d, %s},\n", r, c.origCCC, c.nLeadingNonStarters, c.nTrailingNonStarters, s)
last = current
}
}
fmt.Fprintln(w, "}")
gen.WriteVersionedGoFile("data_test.go", "norm", w.Bytes())
}

117
vendor/golang.org/x/text/unicode/norm/triegen.go generated vendored
View File

@ -1,117 +0,0 @@
// 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.
// +build ignore
// Trie table generator.
// Used by make*tables tools to generate a go file with trie data structures
// for mapping UTF-8 to a 16-bit value. All but the last byte in a UTF-8 byte
// sequence are used to lookup offsets in the index table to be used for the
// next byte. The last byte is used to index into a table with 16-bit values.
package main
import (
"fmt"
"io"
)
const maxSparseEntries = 16
type normCompacter struct {
sparseBlocks [][]uint64
sparseOffset []uint16
sparseCount int
name string
}
func mostFrequentStride(a []uint64) int {
counts := make(map[int]int)
var v int
for _, x := range a {
if stride := int(x) - v; v != 0 && stride >= 0 {
counts[stride]++
}
v = int(x)
}
var maxs, maxc int
for stride, cnt := range counts {
if cnt > maxc || (cnt == maxc && stride < maxs) {
maxs, maxc = stride, cnt
}
}
return maxs
}
func countSparseEntries(a []uint64) int {
stride := mostFrequentStride(a)
var v, count int
for _, tv := range a {
if int(tv)-v != stride {
if tv != 0 {
count++
}
}
v = int(tv)
}
return count
}
func (c *normCompacter) Size(v []uint64) (sz int, ok bool) {
if n := countSparseEntries(v); n <= maxSparseEntries {
return (n+1)*4 + 2, true
}
return 0, false
}
func (c *normCompacter) Store(v []uint64) uint32 {
h := uint32(len(c.sparseOffset))
c.sparseBlocks = append(c.sparseBlocks, v)
c.sparseOffset = append(c.sparseOffset, uint16(c.sparseCount))
c.sparseCount += countSparseEntries(v) + 1
return h
}
func (c *normCompacter) Handler() string {
return c.name + "Sparse.lookup"
}
func (c *normCompacter) Print(w io.Writer) (retErr error) {
p := func(f string, x ...interface{}) {
if _, err := fmt.Fprintf(w, f, x...); retErr == nil && err != nil {
retErr = err
}
}
ls := len(c.sparseBlocks)
p("// %sSparseOffset: %d entries, %d bytes\n", c.name, ls, ls*2)
p("var %sSparseOffset = %#v\n\n", c.name, c.sparseOffset)
ns := c.sparseCount
p("// %sSparseValues: %d entries, %d bytes\n", c.name, ns, ns*4)
p("var %sSparseValues = [%d]valueRange {", c.name, ns)
for i, b := range c.sparseBlocks {
p("\n// Block %#x, offset %#x", i, c.sparseOffset[i])
var v int
stride := mostFrequentStride(b)
n := countSparseEntries(b)
p("\n{value:%#04x,lo:%#02x},", stride, uint8(n))
for i, nv := range b {
if int(nv)-v != stride {
if v != 0 {
p(",hi:%#02x},", 0x80+i-1)
}
if nv != 0 {
p("\n{value:%#04x,lo:%#02x", nv, 0x80+i)
}
}
v = int(nv)
}
if v != 0 {
p(",hi:%#02x},", 0x80+len(b)-1)
}
}
p("\n}\n\n")
return
}

115
vendor/golang.org/x/text/unicode/rangetable/gen.go generated vendored
View File

@ -1,115 +0,0 @@
// Copyright 2015 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
package main
import (
"bytes"
"flag"
"fmt"
"io"
"log"
"reflect"
"strings"
"unicode"
"golang.org/x/text/collate"
"golang.org/x/text/internal/gen"
"golang.org/x/text/internal/ucd"
"golang.org/x/text/language"
"golang.org/x/text/unicode/rangetable"
)
var versionList = flag.String("versions", "",
"list of versions for which to generate RangeTables")
const bootstrapMessage = `No versions specified.
To bootstrap the code generation, run:
go run gen.go --versions=4.1.0,5.0.0,6.0.0,6.1.0,6.2.0,6.3.0,7.0.0
and ensure that the latest versions are included by checking:
https://www.unicode.org/Public/`
func getVersions() []string {
if *versionList == "" {
log.Fatal(bootstrapMessage)
}
c := collate.New(language.Und, collate.Numeric)
versions := strings.Split(*versionList, ",")
c.SortStrings(versions)
// Ensure that at least the current version is included.
for _, v := range versions {
if v == gen.UnicodeVersion() {
return versions
}
}
versions = append(versions, gen.UnicodeVersion())
c.SortStrings(versions)
return versions
}
func main() {
gen.Init()
versions := getVersions()
w := &bytes.Buffer{}
fmt.Fprintf(w, "//go:generate go run gen.go --versions=%s\n\n", strings.Join(versions, ","))
fmt.Fprintf(w, "import \"unicode\"\n\n")
vstr := func(s string) string { return strings.Replace(s, ".", "_", -1) }
fmt.Fprintf(w, "var assigned = map[string]*unicode.RangeTable{\n")
for _, v := range versions {
fmt.Fprintf(w, "\t%q: assigned%s,\n", v, vstr(v))
}
fmt.Fprintf(w, "}\n\n")
var size int
for _, v := range versions {
assigned := []rune{}
r := gen.Open("https://www.unicode.org/Public/", "", v+"/ucd/UnicodeData.txt")
ucd.Parse(r, func(p *ucd.Parser) {
assigned = append(assigned, p.Rune(0))
})
rt := rangetable.New(assigned...)
sz := int(reflect.TypeOf(unicode.RangeTable{}).Size())
sz += int(reflect.TypeOf(unicode.Range16{}).Size()) * len(rt.R16)
sz += int(reflect.TypeOf(unicode.Range32{}).Size()) * len(rt.R32)
fmt.Fprintf(w, "// size %d bytes (%d KiB)\n", sz, sz/1024)
fmt.Fprintf(w, "var assigned%s = ", vstr(v))
print(w, rt)
size += sz
}
fmt.Fprintf(w, "// Total size %d bytes (%d KiB)\n", size, size/1024)
gen.WriteVersionedGoFile("tables.go", "rangetable", w.Bytes())
}
func print(w io.Writer, rt *unicode.RangeTable) {
fmt.Fprintln(w, "&unicode.RangeTable{")
fmt.Fprintln(w, "\tR16: []unicode.Range16{")
for _, r := range rt.R16 {
fmt.Fprintf(w, "\t\t{%#04x, %#04x, %d},\n", r.Lo, r.Hi, r.Stride)
}
fmt.Fprintln(w, "\t},")
fmt.Fprintln(w, "\tR32: []unicode.Range32{")
for _, r := range rt.R32 {
fmt.Fprintf(w, "\t\t{%#08x, %#08x, %d},\n", r.Lo, r.Hi, r.Stride)
}
fmt.Fprintln(w, "\t},")
fmt.Fprintf(w, "\tLatinOffset: %d,\n", rt.LatinOffset)
fmt.Fprintf(w, "}\n\n")
}

260
vendor/golang.org/x/text/unicode/rangetable/merge.go generated vendored
View File

@ -1,260 +0,0 @@
// Copyright 2015 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 rangetable
import (
"unicode"
)
// atEnd is used to mark a completed iteration.
const atEnd = unicode.MaxRune + 1
// Merge returns a new RangeTable that is the union of the given tables.
// It can also be used to compact user-created RangeTables. The entries in
// R16 and R32 for any given RangeTable should be sorted and non-overlapping.
//
// A lookup in the resulting table can be several times faster than using In
// directly on the ranges. Merge is an expensive operation, however, and only
// makes sense if one intends to use the result for more than a couple of
// hundred lookups.
func Merge(ranges ...*unicode.RangeTable) *unicode.RangeTable {
rt := &unicode.RangeTable{}
if len(ranges) == 0 {
return rt
}
iter := tablesIter(make([]tableIndex, len(ranges)))
for i, t := range ranges {
iter[i] = tableIndex{t, 0, atEnd}
if len(t.R16) > 0 {
iter[i].next = rune(t.R16[0].Lo)
}
}
if r0 := iter.next16(); r0.Stride != 0 {
for {
r1 := iter.next16()
if r1.Stride == 0 {
rt.R16 = append(rt.R16, r0)
break
}
stride := r1.Lo - r0.Hi
if (r1.Lo == r1.Hi || stride == r1.Stride) && (r0.Lo == r0.Hi || stride == r0.Stride) {
// Fully merge the next range into the previous one.
r0.Hi, r0.Stride = r1.Hi, stride
continue
} else if stride == r0.Stride {
// Move the first element of r1 to r0. This may eliminate an
// entry.
r0.Hi = r1.Lo
r0.Stride = stride
r1.Lo = r1.Lo + r1.Stride
if r1.Lo > r1.Hi {
continue
}
}
rt.R16 = append(rt.R16, r0)
r0 = r1
}
}
for i, t := range ranges {
iter[i] = tableIndex{t, 0, atEnd}
if len(t.R32) > 0 {
iter[i].next = rune(t.R32[0].Lo)
}
}
if r0 := iter.next32(); r0.Stride != 0 {
for {
r1 := iter.next32()
if r1.Stride == 0 {
rt.R32 = append(rt.R32, r0)
break
}
stride := r1.Lo - r0.Hi
if (r1.Lo == r1.Hi || stride == r1.Stride) && (r0.Lo == r0.Hi || stride == r0.Stride) {
// Fully merge the next range into the previous one.
r0.Hi, r0.Stride = r1.Hi, stride
continue
} else if stride == r0.Stride {
// Move the first element of r1 to r0. This may eliminate an
// entry.
r0.Hi = r1.Lo
r1.Lo = r1.Lo + r1.Stride
if r1.Lo > r1.Hi {
continue
}
}
rt.R32 = append(rt.R32, r0)
r0 = r1
}
}
for i := 0; i < len(rt.R16) && rt.R16[i].Hi <= unicode.MaxLatin1; i++ {
rt.LatinOffset = i + 1
}
return rt
}
type tableIndex struct {
t *unicode.RangeTable
p uint32
next rune
}
type tablesIter []tableIndex
// sortIter does an insertion sort using the next field of tableIndex. Insertion
// sort is a good sorting algorithm for this case.
func sortIter(t []tableIndex) {
for i := range t {
for j := i; j > 0 && t[j-1].next > t[j].next; j-- {
t[j], t[j-1] = t[j-1], t[j]
}
}
}
// next16 finds the ranged to be added to the table. If ranges overlap between
// multiple tables it clips the result to a non-overlapping range if the
// elements are not fully subsumed. It returns a zero range if there are no more
// ranges.
func (ti tablesIter) next16() unicode.Range16 {
sortIter(ti)
t0 := ti[0]
if t0.next == atEnd {
return unicode.Range16{}
}
r0 := t0.t.R16[t0.p]
r0.Lo = uint16(t0.next)
// We restrict the Hi of the current range if it overlaps with another range.
for i := range ti {
tn := ti[i]
// Since our tableIndices are sorted by next, we can break if the there
// is no overlap. The first value of a next range can always be merged
// into the current one, so we can break in case of equality as well.
if rune(r0.Hi) <= tn.next {
break
}
rn := tn.t.R16[tn.p]
rn.Lo = uint16(tn.next)
// Limit r0.Hi based on next ranges in list, but allow it to overlap
// with ranges as long as it subsumes it.
m := (rn.Lo - r0.Lo) % r0.Stride
if m == 0 && (rn.Stride == r0.Stride || rn.Lo == rn.Hi) {
// Overlap, take the min of the two Hi values: for simplicity's sake
// we only process one range at a time.
if r0.Hi > rn.Hi {
r0.Hi = rn.Hi
}
} else {
// Not a compatible stride. Set to the last possible value before
// rn.Lo, but ensure there is at least one value.
if x := rn.Lo - m; r0.Lo <= x {
r0.Hi = x
}
break
}
}
// Update the next values for each table.
for i := range ti {
tn := &ti[i]
if rune(r0.Hi) < tn.next {
break
}
rn := tn.t.R16[tn.p]
stride := rune(rn.Stride)
tn.next += stride * (1 + ((rune(r0.Hi) - tn.next) / stride))
if rune(rn.Hi) < tn.next {
if tn.p++; int(tn.p) == len(tn.t.R16) {
tn.next = atEnd
} else {
tn.next = rune(tn.t.R16[tn.p].Lo)
}
}
}
if r0.Lo == r0.Hi {
r0.Stride = 1
}
return r0
}
// next32 finds the ranged to be added to the table. If ranges overlap between
// multiple tables it clips the result to a non-overlapping range if the
// elements are not fully subsumed. It returns a zero range if there are no more
// ranges.
func (ti tablesIter) next32() unicode.Range32 {
sortIter(ti)
t0 := ti[0]
if t0.next == atEnd {
return unicode.Range32{}
}
r0 := t0.t.R32[t0.p]
r0.Lo = uint32(t0.next)
// We restrict the Hi of the current range if it overlaps with another range.
for i := range ti {
tn := ti[i]
// Since our tableIndices are sorted by next, we can break if the there
// is no overlap. The first value of a next range can always be merged
// into the current one, so we can break in case of equality as well.
if rune(r0.Hi) <= tn.next {
break
}
rn := tn.t.R32[tn.p]
rn.Lo = uint32(tn.next)
// Limit r0.Hi based on next ranges in list, but allow it to overlap
// with ranges as long as it subsumes it.
m := (rn.Lo - r0.Lo) % r0.Stride
if m == 0 && (rn.Stride == r0.Stride || rn.Lo == rn.Hi) {
// Overlap, take the min of the two Hi values: for simplicity's sake
// we only process one range at a time.
if r0.Hi > rn.Hi {
r0.Hi = rn.Hi
}
} else {
// Not a compatible stride. Set to the last possible value before
// rn.Lo, but ensure there is at least one value.
if x := rn.Lo - m; r0.Lo <= x {
r0.Hi = x
}
break
}
}
// Update the next values for each table.
for i := range ti {
tn := &ti[i]
if rune(r0.Hi) < tn.next {
break
}
rn := tn.t.R32[tn.p]
stride := rune(rn.Stride)
tn.next += stride * (1 + ((rune(r0.Hi) - tn.next) / stride))
if rune(rn.Hi) < tn.next {
if tn.p++; int(tn.p) == len(tn.t.R32) {
tn.next = atEnd
} else {
tn.next = rune(tn.t.R32[tn.p].Lo)
}
}
}
if r0.Lo == r0.Hi {
r0.Stride = 1
}
return r0
}

70
vendor/golang.org/x/text/unicode/rangetable/rangetable.go generated vendored
View File

@ -1,70 +0,0 @@
// Copyright 2015 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 rangetable provides utilities for creating and inspecting
// unicode.RangeTables.
package rangetable
import (
"sort"
"unicode"
)
// New creates a RangeTable from the given runes, which may contain duplicates.
func New(r ...rune) *unicode.RangeTable {
if len(r) == 0 {
return &unicode.RangeTable{}
}
sort.Sort(byRune(r))
// Remove duplicates.
k := 1
for i := 1; i < len(r); i++ {
if r[k-1] != r[i] {
r[k] = r[i]
k++
}
}
var rt unicode.RangeTable
for _, r := range r[:k] {
if r <= 0xFFFF {
rt.R16 = append(rt.R16, unicode.Range16{Lo: uint16(r), Hi: uint16(r), Stride: 1})
} else {
rt.R32 = append(rt.R32, unicode.Range32{Lo: uint32(r), Hi: uint32(r), Stride: 1})
}
}
// Optimize RangeTable.
return Merge(&rt)
}
type byRune []rune
func (r byRune) Len() int { return len(r) }
func (r byRune) Swap(i, j int) { r[i], r[j] = r[j], r[i] }
func (r byRune) Less(i, j int) bool { return r[i] < r[j] }
// Visit visits all runes in the given RangeTable in order, calling fn for each.
func Visit(rt *unicode.RangeTable, fn func(rune)) {
for _, r16 := range rt.R16 {
for r := rune(r16.Lo); r <= rune(r16.Hi); r += rune(r16.Stride) {
fn(r)
}
}
for _, r32 := range rt.R32 {
for r := rune(r32.Lo); r <= rune(r32.Hi); r += rune(r32.Stride) {
fn(r)
}
}
}
// Assigned returns a RangeTable with all assigned code points for a given
// Unicode version. This includes graphic, format, control, and private-use
// characters. It returns nil if the data for the given version is not
// available.
func Assigned(version string) *unicode.RangeTable {
return assigned[version]
}

6378
vendor/golang.org/x/text/unicode/rangetable/tables10.0.0.go generated vendored
View File

File diff suppressed because it is too large Load Diff

7029
vendor/golang.org/x/text/unicode/rangetable/tables11.0.0.go generated vendored
View File

File diff suppressed because it is too large Load Diff

5737
vendor/golang.org/x/text/unicode/rangetable/tables9.0.0.go generated vendored
View File

File diff suppressed because it is too large Load Diff