An implementation of grapheme cluster boundaries from Unicode text segmentation (UAX 29), for Unicode 17.
Quick start
go get "github.com/clipperhouse/uax29/v2/graphemes"
import "github.com/clipperhouse/uax29/v2/graphemes"
text := "Hello, 世界. Nice dog! 👍🐶"
tokens := graphemes.FromString(text)
for tokens.Next() { // Next() returns true until end of data
fmt.Println(tokens.Value()) // Do something with the current grapheme
}
A grapheme is a “single visible character”, which might be a simple as a single letter, or a complex emoji that consists of several Unicode code points.
Conformance
We use the Unicode test suite.
APIs
If you have a string
text := "Hello, 世界. Nice dog! 👍🐶"
tokens := graphemes.FromString(text)
for tokens.Next() { // Next() returns true until end of data
fmt.Println(tokens.Value()) // Do something with the current grapheme
}
If you have an io.Reader
FromReader embeds a bufio.Scanner, so just use those methods.
r := getYourReader() // from a file or network maybe
tokens := graphemes.FromReader(r)
for tokens.Scan() { // Scan() returns true until error or EOF
fmt.Println(tokens.Text()) // Do something with the current grapheme
}
if tokens.Err() != nil { // Check the error
log.Fatal(tokens.Err())
}
If you have a []byte
b := []byte("Hello, 世界. Nice dog! 👍🐶")
tokens := graphemes.FromBytes(b)
for tokens.Next() { // Next() returns true until end of data
fmt.Println(tokens.Value()) // Do something with the current grapheme
}
Benchmarks
goos: darwin
goarch: arm64
pkg: github.com/clipperhouse/uax29/graphemes/comparative
cpu: Apple M2
BenchmarkGraphemesMixed/clipperhouse/uax29-8 142635 ns/op 245.12 MB/s 0 B/op 0 allocs/op
BenchmarkGraphemesMixed/rivo/uniseg-8 2018284 ns/op 17.32 MB/s 0 B/op 0 allocs/op
BenchmarkGraphemesASCII/clipperhouse/uax29-8 8846 ns/op 508.73 MB/s 0 B/op 0 allocs/op
BenchmarkGraphemesASCII/rivo/uniseg-8 366760 ns/op 12.27 MB/s 0 B/op 0 allocs/op
Invalid inputs
Invalid UTF-8 input is considered undefined behavior. We test to ensure that bad inputs will not cause pathological outcomes, such as a panic or infinite loop. Callers should expect “garbage-in, garbage-out”.
Your pipeline should probably include a call to utf8.Valid().