diff options
Diffstat (limited to 'vendor/github.com/couchbase/vellum/utf8/utf8.go')
| -rw-r--r-- | vendor/github.com/couchbase/vellum/utf8/utf8.go | 98 |
1 files changed, 60 insertions, 38 deletions
diff --git a/vendor/github.com/couchbase/vellum/utf8/utf8.go b/vendor/github.com/couchbase/vellum/utf8/utf8.go index 47dbe9d1c5..54e23b937c 100644 --- a/vendor/github.com/couchbase/vellum/utf8/utf8.go +++ b/vendor/github.com/couchbase/vellum/utf8/utf8.go @@ -25,19 +25,39 @@ type Sequences []Sequence // NewSequences constructs a collection of Sequence which describe the // byte ranges covered between the start and end runes. func NewSequences(start, end rune) (Sequences, error) { - var rv Sequences + rv, _, err := NewSequencesPrealloc(start, end, nil, nil, nil, nil) + return rv, err +} + +func NewSequencesPrealloc(start, end rune, + preallocSequences Sequences, + preallocRangeStack RangeStack, + preallocStartBytes, preallocEndBytes []byte) (Sequences, RangeStack, error) { + rv := preallocSequences[:0] + + startBytes := preallocStartBytes + if cap(startBytes) < utf8.UTFMax { + startBytes = make([]byte, utf8.UTFMax) + } + startBytes = startBytes[:utf8.UTFMax] - var rangeStack rangeStack - rangeStack = rangeStack.Push(&scalarRange{start, end}) + endBytes := preallocEndBytes + if cap(endBytes) < utf8.UTFMax { + endBytes = make([]byte, utf8.UTFMax) + } + endBytes = endBytes[:utf8.UTFMax] + + rangeStack := preallocRangeStack[:0] + rangeStack = rangeStack.Push(scalarRange{start, end}) rangeStack, r := rangeStack.Pop() TOP: - for r != nil { + for r != nilScalarRange { INNER: for { r1, r2 := r.split() - if r1 != nil { - rangeStack = rangeStack.Push(&scalarRange{r2.start, r2.end}) + if r1 != nilScalarRange { + rangeStack = rangeStack.Push(scalarRange{r2.start, r2.end}) r.start = r1.start r.end = r1.end continue INNER @@ -49,13 +69,13 @@ TOP: for i := 1; i < utf8.UTFMax; i++ { max := maxScalarValue(i) if r.start <= max && max < r.end { - rangeStack = rangeStack.Push(&scalarRange{max + 1, r.end}) + rangeStack = rangeStack.Push(scalarRange{max + 1, r.end}) r.end = max continue INNER } } asciiRange := r.ascii() - if asciiRange != nil { + if asciiRange != nilRange { rv = append(rv, Sequence{ asciiRange, }) @@ -66,23 +86,21 @@ TOP: m := rune((1 << (6 * i)) - 1) if (r.start & ^m) != (r.end & ^m) { if (r.start & m) != 0 { - rangeStack = rangeStack.Push(&scalarRange{(r.start | m) + 1, r.end}) + rangeStack = rangeStack.Push(scalarRange{(r.start | m) + 1, r.end}) r.end = r.start | m continue INNER } if (r.end & m) != m { - rangeStack = rangeStack.Push(&scalarRange{r.end & ^m, r.end}) + rangeStack = rangeStack.Push(scalarRange{r.end & ^m, r.end}) r.end = (r.end & ^m) - 1 continue INNER } } } - start := make([]byte, utf8.UTFMax) - end := make([]byte, utf8.UTFMax) - n, m := r.encode(start, end) - seq, err := SequenceFromEncodedRange(start[0:n], end[0:m]) + n, m := r.encode(startBytes, endBytes) + seq, err := SequenceFromEncodedRange(startBytes[0:n], endBytes[0:m]) if err != nil { - return nil, err + return nil, nil, err } rv = append(rv, seq) rangeStack, r = rangeStack.Pop() @@ -90,11 +108,11 @@ TOP: } } - return rv, nil + return rv, rangeStack, nil } -// Sequence is a collection of *Range -type Sequence []*Range +// Sequence is a collection of Range +type Sequence []Range // SequenceFromEncodedRange creates sequence from the encoded bytes func SequenceFromEncodedRange(start, end []byte) (Sequence, error) { @@ -104,21 +122,21 @@ func SequenceFromEncodedRange(start, end []byte) (Sequence, error) { switch len(start) { case 2: return Sequence{ - &Range{start[0], end[0]}, - &Range{start[1], end[1]}, + Range{start[0], end[0]}, + Range{start[1], end[1]}, }, nil case 3: return Sequence{ - &Range{start[0], end[0]}, - &Range{start[1], end[1]}, - &Range{start[2], end[2]}, + Range{start[0], end[0]}, + Range{start[1], end[1]}, + Range{start[2], end[2]}, }, nil case 4: return Sequence{ - &Range{start[0], end[0]}, - &Range{start[1], end[1]}, - &Range{start[2], end[2]}, - &Range{start[3], end[3]}, + Range{start[0], end[0]}, + Range{start[1], end[1]}, + Range{start[2], end[2]}, + Range{start[3], end[3]}, }, nil } @@ -159,6 +177,8 @@ type Range struct { End byte } +var nilRange = Range{0xff, 0} + func (u Range) matches(b byte) bool { if u.Start <= b && b <= u.End { return true @@ -178,37 +198,39 @@ type scalarRange struct { end rune } +var nilScalarRange = scalarRange{0xffff, 0} + func (s *scalarRange) String() string { return fmt.Sprintf("ScalarRange(%d,%d)", s.start, s.end) } // split this scalar range if it overlaps with a surrogate codepoint -func (s *scalarRange) split() (*scalarRange, *scalarRange) { +func (s *scalarRange) split() (scalarRange, scalarRange) { if s.start < 0xe000 && s.end > 0xd7ff { - return &scalarRange{ + return scalarRange{ start: s.start, end: 0xd7ff, }, - &scalarRange{ + scalarRange{ start: 0xe000, end: s.end, } } - return nil, nil + return nilScalarRange, nilScalarRange } func (s *scalarRange) valid() bool { return s.start <= s.end } -func (s *scalarRange) ascii() *Range { +func (s *scalarRange) ascii() Range { if s.valid() && s.end <= 0x7f { - return &Range{ + return Range{ Start: byte(s.start), End: byte(s.end), } } - return nil + return nilRange } // start and end MUST have capacity for utf8.UTFMax bytes @@ -218,16 +240,16 @@ func (s *scalarRange) encode(start, end []byte) (int, int) { return n, m } -type rangeStack []*scalarRange +type RangeStack []scalarRange -func (s rangeStack) Push(v *scalarRange) rangeStack { +func (s RangeStack) Push(v scalarRange) RangeStack { return append(s, v) } -func (s rangeStack) Pop() (rangeStack, *scalarRange) { +func (s RangeStack) Pop() (RangeStack, scalarRange) { l := len(s) if l < 1 { - return s, nil + return s, nilScalarRange } return s[:l-1], s[l-1] } |