// 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)
}
func appendRange32(t *unicode.RangeTable, lo, hi uint32) error {
if t.R32 == nil || len(t.R32) == 0 {
t.R32 = append(t.R32, unicode.Range32{lo, hi, 1})
return nil
}
range_ := &t.R32[len(t.R32)-1]
if lo <= range_.Hi {
return fmt.Errorf("AppendRange: lo must be greater than Hi of the last range: lo=%d", lo)
}
if lo == hi {
if range_.Hi+range_.Stride == lo {
range_.Hi = lo
if len(t.R32) >= 2 {
prevR := &t.R32[len(t.R32)-2]
if prevR.Stride > 1 && prevR.Hi+1 == range_.Lo {
prevCount := (prevR.Hi-prevR.Lo)/prevR.Stride + 1
count := (range_.Hi-range_.Lo)/range_.Stride + 1
if prevCount <= count {
range_.Lo--
prevR.Hi -= prevR.Stride
if prevR.Lo == prevR.Hi {
prevR.Stride = 1
}
}
}
}
} else if range_.Lo == range_.Hi {
range_.Hi = lo
range_.Stride = lo - range_.Lo
} else {
t.R32 = append(t.R32, unicode.Range32{lo, hi, 1})
}
} else {
if range_.Stride == 1 {
if lo-1 <= range_.Hi {
range_.Hi = hi
} else {
t.R32 = append(t.R32, unicode.Range32{lo, hi, 1})
}
} else {
if lo-1 <= range_.Hi {
range_.Hi -= range_.Stride
if range_.Lo == range_.Hi {
range_.Stride = 1
}
t.R32 = append(t.R32, unicode.Range32{lo - 1, hi, 1})
} else {
t.R32 = append(t.R32, unicode.Range32{lo, hi, 1})
}
}
}
return nil
}
func addR16ToTable(r *unicode.RangeTable, r16 unicode.Range16) {
if r.R16 == nil {
r.R16 = make([]unicode.Range16, 0, 1)
}
r.R16 = append(r.R16, r16)
if r16.Hi <= unicode.MaxLatin1 {
r.LatinOffset++
}
}
func CharClass(negated bool, table *unicode.RangeTable) Matcher {
var t unicode.RangeTable
t.R16 = make([]unicode.Range16, len(table.R16))
copy(t.R16, table.R16)
t.R32 = make([]unicode.Range32, len(table.R32))
copy(t.R32, table.R32)
t.LatinOffset = table.LatinOffset
return charClassMatcher{negated, t}
}
func OneOf(chars ...rune) Matcher {
if len(chars) == 0 {
return Fail()
}
var t unicode.RangeTable
sort.Sort(utf8betical(chars))
for _, ch := range chars {
if ch >= 0x10000 {
t.R32 = append(t.R32, unicode.Range32{Lo: uint32(ch), Hi: uint32(ch), Stride: 1})
} else {
t.R16 = append(t.R16, unicode.Range16{Lo: uint16(ch), Hi: uint16(ch), Stride: 1})
if ch < 0x100 {
t.LatinOffset += 1
}
}
}
return charClassMatcher{false, t}
}
func Range(lo, hi rune) Matcher {
if lo > hi {
return Fail()
}
var t unicode.RangeTable
if lo >= 0x10000 {
t.R32 = append(t.R32, unicode.Range32{Lo: uint32(lo), Hi: uint32(hi), Stride: 1})
} else if hi < 0x10000 {
t.R16 = append(t.R16, unicode.Range16{Lo: uint16(lo), Hi: uint16(hi), Stride: 1})
if hi < 0x100 {
t.LatinOffset = 1
}
} else {
t.R16 = append(t.R16, unicode.Range16{Lo: uint16(lo), Hi: 0xffff, Stride: 1})
t.R32 = append(t.R32, unicode.Range32{Lo: 0x10000, Hi: uint32(hi), Stride: 1})
}
return charClassMatcher{false, t}
}
func appendRange16(t *unicode.RangeTable, lo, hi uint16) error {
if t.R16 == nil || len(t.R16) == 0 {
t.R16 = append(t.R16, unicode.Range16{lo, hi, 1})
if hi <= unicode.MaxLatin1 {
t.LatinOffset++
}
return nil
}
range_ := &t.R16[len(t.R16)-1]
if lo <= range_.Hi {
return fmt.Errorf("AppendRange: lo must be greater than Hi of the last range: lo=%d", lo)
}
if lo == hi {
if range_.Hi+range_.Stride == lo {
range_.Hi = lo
if len(t.R16) >= 2 {
prevR := &t.R16[len(t.R16)-2]
if prevR.Stride > 1 && prevR.Hi+1 == range_.Lo {
prevCount := (prevR.Hi-prevR.Lo)/prevR.Stride + 1
count := (range_.Hi-range_.Lo)/range_.Stride + 1
if prevCount <= count {
range_.Lo--
prevR.Hi -= prevR.Stride
if prevR.Lo == prevR.Hi {
prevR.Stride = 1
}
}
}
}
} else if range_.Lo == range_.Hi {
range_.Hi = lo
range_.Stride = lo - range_.Lo
} else {
t.R16 = append(t.R16, unicode.Range16{lo, hi, 1})
if hi <= unicode.MaxLatin1 {
t.LatinOffset++
}
}
} else {
if range_.Stride == 1 {
if lo-1 <= range_.Hi {
range_.Hi = hi
} else {
if range_.Lo == range_.Hi {
range_.Hi = hi
range_.Stride = range_.Hi - range_.Lo
} else {
t.R16 = append(t.R16, unicode.Range16{lo, hi, 1})
if hi <= unicode.MaxLatin1 {
t.LatinOffset++
}
}
}
} else {
if lo-1 <= range_.Hi {
range_.Hi -= range_.Stride
if range_.Lo == range_.Hi {
range_.Stride = 1
}
t.R16 = append(t.R16, unicode.Range16{lo - 1, hi, 1})
if hi <= unicode.MaxLatin1 {
t.LatinOffset++
}
} else {
t.R16 = append(t.R16, unicode.Range16{lo, hi, 1})
if hi <= unicode.MaxLatin1 {
t.LatinOffset++
}
}
}
}
return nil
}
func (n *Nfa) parseTerm(s *ScannerSource, in0 *NfaState, nest int) (in, out *NfaState) {
if in = in0; in == nil {
in = n.NewState()
}
switch arune := s.Current(); arune {
default:
s.Move()
out = in.AddConsuming(NewRuneEdge(n.NewState(), arune)).Target()
case '+', '*', '?':
panic(fmt.Errorf("unexpected metachar %q", string(arune)))
case '\\':
switch arune = s.mustParseChar("ApPz"); arune {
default:
out = in.AddConsuming(NewRuneEdge(n.NewState(), arune)).Target()
case 'A':
out = in.AddNonConsuming(NewAssertEdge(n.NewState(), TextStart)).Target()
case 'z':
out = in.AddNonConsuming(NewAssertEdge(n.NewState(), TextEnd)).Target()
case 'p', 'P':
name, ok := "", false
s.expect('{')
for !s.Accept('}') {
name += string(s.mustGetChar())
}
var ranges *unicode.RangeTable
if ranges, ok = unicode.Categories[name]; !ok {
if ranges, ok = unicode.Scripts[name]; !ok {
panic(fmt.Errorf("unknown Unicode category name %q", name))
}
}
out = in.AddConsuming(NewRangesEdge(n.NewState(), arune == 'P', ranges)).Target()
}
case 0, '|':
return nil, nil
case ')':
if nest == 0 {
panic(fmt.Errorf(`unexpected ")"`))
}
return nil, nil
case '(':
s.Move()
in, out = n.parseExpr(s, in, n.NewState(), nest+1)
s.expect(')')
case '.': // All but '\U+0000', '\n'
s.Move()
out = in.AddConsuming(NewRangesEdge(n.NewState(), true, &unicode.RangeTable{R16: []unicode.Range16{{'\n', '\n', 1}}})).Target()
case '^':
s.Move()
out = in.AddNonConsuming(NewAssertEdge(n.NewState(), LineStart)).Target()
case '$':
s.Move()
out = in.AddNonConsuming(NewAssertEdge(n.NewState(), LineEnd)).Target()
case '[':
s.Move()
ranges := &unicode.RangeTable{}
invert := s.Accept('^')
loop:
for {
a := s.mustParseChar("-")
switch s.Current() {
case '\\':
ranges.R32 = append(ranges.R32, unicode.Range32{uint32(a), uint32(a), 1})
a := s.mustParseChar("-")
ranges.R32 = append(ranges.R32, unicode.Range32{uint32(a), uint32(a), 1})
default:
if s.Accept('-') {
// Allow `[+-]`
if s.Current() == ']' {
s.Move()
ranges.R32 = append(ranges.R32, unicode.Range32{uint32(a), uint32(a), 1})
ranges.R32 = append(ranges.R32, unicode.Range32{'-', '-', 1})
break loop
}
b := s.mustParseChar("")
if b < a {
panic(fmt.Errorf(`missing or invalid range bounds ordering in bracket expression "%s-%s"`, string(a), string(b)))
}
ranges.R32 = append(ranges.R32, unicode.Range32{uint32(a), uint32(b), 1})
} else {
ranges.R32 = append(ranges.R32, unicode.Range32{uint32(a), uint32(a), 1})
}
}
if s.Accept(']') {
break
}
}
(*rangeSlice)(&ranges.R32).normalize()
out = in.AddConsuming(NewRangesEdge(n.NewState(), invert, ranges)).Target()
}
// postfix ops
switch s.Current() {
case '+':
s.Move()
_, out = n.OneOrMore(in, out)
case '*':
s.Move()
_, out = n.ZeroOrMore(in, out)
case '?':
s.Move()
_, out = n.ZeroOrOne(in, out)
}
return
}
func addR32ToTable(r *unicode.RangeTable, r32 unicode.Range32) {
if r.R32 == nil {
r.R32 = make([]unicode.Range32, 0, 1)
}
r.R32 = append(r.R32, r32)
}
