Use SIMD to replace C0 control codes in Go code

This commit is contained in:
Kovid Goyal
2025-07-21 08:54:22 +05:30
parent 12c1b0cbdf
commit fd5876b94e
10 changed files with 74 additions and 47 deletions

View File

@@ -14,6 +14,7 @@ import (
"strings"
"sync"
"github.com/kovidgoyal/kitty/tools/simdstring"
"golang.org/x/exp/constraints"
"golang.org/x/text/language"
)
@@ -423,3 +424,35 @@ var LanguageTag = sync.OnceValue(func() language.Tag {
return tag
})
// Replace control codes by unicode codepoints that describe the codes
// making the text safe to send to a terminal
func ReplaceControlCodes(text, replace_tab_by, replace_newline_by string) string {
buf := strings.Builder{}
for len(text) > 0 {
idx := simdstring.IndexC0String(text)
if idx < 0 {
if buf.Cap() == 0 {
return text
}
buf.WriteString(text)
break
}
if buf.Cap() == 0 {
buf.Grow(2 * len(text))
}
buf.WriteString(text[:idx])
switch text[idx] {
case '\n':
buf.WriteString(replace_newline_by)
case '\t':
buf.WriteString(replace_tab_by)
case 0x7f:
buf.WriteRune(0x2421)
default:
buf.WriteRune(0x2400 + rune(text[idx]))
}
text = text[idx+1:]
}
return buf.String()
}

View File

@@ -34,3 +34,17 @@ func TestStringScanner(t *testing.T) {
}
}
}
func TestReplaceControlCodes(t *testing.T) {
for text, expected := range map[string]string{
"none": "none",
"a\r\x01b\x03\x7f c\n\td": "a\u240d\u2401b\u2403\u2421 cX d",
"\x01": "\u2401",
"\x00\x0b": "\u2400\u240b",
} {
actual := ReplaceControlCodes(text, " ", "X")
if diff := cmp.Diff(expected, actual); diff != "" {
t.Fatalf("Failed for text: %#v\n%s", text, diff)
}
}
}