I just want to find 3-byte character in Go using regexp. But it panic with
regexp: Compile(
\x{E29AA4}): error parsing regexp: invalid escape sequence:\x{E29AA4
func get_words_from(text string) []string {
words := regexp.MustCompile(`\x{E29AA4}`)
return words.FindAllString(text, -1)
}
func main() {
text := "One,ВАПОЛтлдо⚤two ыаплд⚤ы ыапю.ы./\tавthree!"
fmt.Println(get_words_from(text))
}
You can try on playground
Decode the UTF-8 byte sequence E2 9A A4 with e.g. utf8.DecodeRune() and use the resulting rune in the regexp:
func get_words_from(text string) []string {
r, _ := utf8.DecodeRune([]byte{0xE2, 0x9A, 0xA4})
words := regexp.MustCompile(string(r))
return words.FindAllString(text, -1)
}
You may also simply convert the byte slice to string (which interprets it as UTF-8 encoded bytes):
func get_words_from2(text string) []string {
s := string([]byte{0xE2, 0x9A, 0xA4})
words := regexp.MustCompile(s)
return words.FindAllString(text, -1)
}
Or use the equivalent unicode code point (which is 0x26a4) in the regexp string:
func get_words_from3(text string) []string {
words := regexp.MustCompile("\u26a4")
return words.FindAllString(text, -1)
}
Note that "\u26a4" is an interpreted string literal and will be unescaped by the Go compiler (not the regexp package).
All examples return (try the examples on the Go Playground):
[⚤ ⚤]
To filter out all runes that have 3 or more bytes in UTF-8, you may use a for range and utf8.RuneLen():
text := "One,ВАПОЛтлдо⚤two ыаплд⚤ы ыапю.ы./\tавthree!"
fmt.Println(text)
var out []rune
for _, r := range text {
if utf8.RuneLen(r) < 3 {
out = append(out, r)
}
}
fmt.Println(string(out))
This outputs (try it on the Go Playground):
One,ВАПОЛтлдо⚤two ыаплд⚤ы ыапю.ы./ авthree!
One,ВАПОЛтлдоtwo ыаплды ыапю.ы./ авthree!
Or use strings.Map(), where you return -1 for such runes, which then will be left out in the result:
out := strings.Map(func(r rune) rune {
if utf8.RuneLen(r) < 3 {
return r
}
return -1
}, text)
fmt.Println(string(out))
This outputs the same. Try this one on the Go Playground.
Also I found that character ⚤ in regex can match by \xE2\x9A\xA4 instead of wrong: \x{E29AA4}