terraformDummyRepo2/vendor/github.com/apparentlymart/go-textseg/textseg/utf8_seqs.go
Malar Invention 00ebcd295e add vendor
2022-04-03 09:37:16 +05:30

19 lines
521 B
Go

package textseg
import "unicode/utf8"
// ScanGraphemeClusters is a split function for bufio.Scanner that splits
// on UTF8 sequence boundaries.
//
// This is included largely for completeness, since this behavior is already
// built in to Go when ranging over a string.
func ScanUTF8Sequences(data []byte, atEOF bool) (int, []byte, error) {
if len(data) == 0 {
return 0, nil, nil
}
r, seqLen := utf8.DecodeRune(data)
if r == utf8.RuneError && !atEOF {
return 0, nil, nil
}
return seqLen, data[:seqLen], nil
}