Support colon delimiter in ANSI escape sequences

# Both should work
    printf "\e[38;5;208mOption 1\e[m\nOption 2" | fzf --ansi
    printf "\e[38:5:208mOption 1\e[m\nOption 2" | fzf --ansi

This change makes ANSI parsing slightly slower.

    cpu: Intel(R) Core(TM) i7-8850H CPU @ 2.60GHz

    Before:
      BenchmarkNextAnsiEscapeSequence-12            992.22 MB/s
      BenchmarkExtractColor-12                      174.35 MB/s

    After:
      BenchmarkNextAnsiEscapeSequence-12            925.05 MB/s
      BenchmarkExtractColor-12                      163.33 MB/s

Fix #2913
This commit is contained in:
Junegunn Choi 2022-08-12 22:18:10 +09:00
parent f4fd53211a
commit aa10dccf90
No known key found for this signature in database
GPG Key ID: 254BC280FEF9C627
3 changed files with 39 additions and 18 deletions

View File

@ -1,6 +1,14 @@
CHANGELOG CHANGELOG
========= =========
0.32.2
------
- ANSI color sequences with colon delimiters are now supported.
```sh
printf "\e[38;5;208mOption 1\e[m\nOption 2" | fzf --ansi
printf "\e[38:5:208mOption 1\e[m\nOption 2" | fzf --ansi
```
0.32.1 0.32.1
------ ------
- Fixed incorrect ordering of `--tiebreak=chunk` - Fixed incorrect ordering of `--tiebreak=chunk`

View File

@ -85,7 +85,7 @@ func isPrint(c uint8) bool {
} }
func matchOperatingSystemCommand(s string) int { func matchOperatingSystemCommand(s string) int {
// `\x1b][0-9];[[:print:]]+(?:\x1b\\\\|\x07)` // `\x1b][0-9][;:][[:print:]]+(?:\x1b\\\\|\x07)`
// ^ match starting here // ^ match starting here
// //
i := 5 // prefix matched in nextAnsiEscapeSequence() i := 5 // prefix matched in nextAnsiEscapeSequence()
@ -103,11 +103,11 @@ func matchOperatingSystemCommand(s string) int {
} }
func matchControlSequence(s string) int { func matchControlSequence(s string) int {
// `\x1b[\\[()][0-9;?]*[a-zA-Z@]` // `\x1b[\\[()][0-9;:?]*[a-zA-Z@]`
// ^ match starting here // ^ match starting here
// //
i := 2 // prefix matched in nextAnsiEscapeSequence() i := 2 // prefix matched in nextAnsiEscapeSequence()
for ; i < len(s) && (isNumeric(s[i]) || s[i] == ';' || s[i] == '?'); i++ { for ; i < len(s) && (isNumeric(s[i]) || s[i] == ';' || s[i] == ':' || s[i] == '?'); i++ {
} }
if i < len(s) { if i < len(s) {
c := s[i] c := s[i]
@ -125,7 +125,7 @@ func isCtrlSeqStart(c uint8) bool {
// nextAnsiEscapeSequence returns the ANSI escape sequence and is equivalent to // nextAnsiEscapeSequence returns the ANSI escape sequence and is equivalent to
// calling FindStringIndex() on the below regex (which was originally used): // calling FindStringIndex() on the below regex (which was originally used):
// //
// "(?:\x1b[\\[()][0-9;?]*[a-zA-Z@]|\x1b][0-9];[[:print:]]+(?:\x1b\\\\|\x07)|\x1b.|[\x0e\x0f]|.\x08)" // "(?:\x1b[\\[()][0-9;:?]*[a-zA-Z@]|\x1b][0-9][;:][[:print:]]+(?:\x1b\\\\|\x07)|\x1b.|[\x0e\x0f]|.\x08)"
func nextAnsiEscapeSequence(s string) (int, int) { func nextAnsiEscapeSequence(s string) (int, int) {
// fast check for ANSI escape sequences // fast check for ANSI escape sequences
i := 0 i := 0
@ -153,16 +153,16 @@ Loop:
return i - n, i + 1 return i - n, i + 1
} }
case '\x1b': case '\x1b':
// match: `\x1b[\\[()][0-9;?]*[a-zA-Z@]` // match: `\x1b[\\[()][0-9;:?]*[a-zA-Z@]`
if i+2 < len(s) && isCtrlSeqStart(s[i+1]) { if i+2 < len(s) && isCtrlSeqStart(s[i+1]) {
if j := matchControlSequence(s[i:]); j != -1 { if j := matchControlSequence(s[i:]); j != -1 {
return i, i + j return i, i + j
} }
} }
// match: `\x1b][0-9];[[:print:]]+(?:\x1b\\\\|\x07)` // match: `\x1b][0-9][;:][[:print:]]+(?:\x1b\\\\|\x07)`
if i+5 < len(s) && s[i+1] == ']' && isNumeric(s[i+2]) && if i+5 < len(s) && s[i+1] == ']' && isNumeric(s[i+2]) &&
s[i+3] == ';' && isPrint(s[i+4]) { (s[i+3] == ';' || s[i+3] == ':') && isPrint(s[i+4]) {
if j := matchOperatingSystemCommand(s[i:]); j != -1 { if j := matchOperatingSystemCommand(s[i:]); j != -1 {
return i, i + j return i, i + j
@ -279,9 +279,20 @@ func extractColor(str string, state *ansiState, proc func(string, *ansiState) bo
return trimmed, nil, state return trimmed, nil, state
} }
func parseAnsiCode(s string) (int, string) { func parseAnsiCode(s string, delimiter byte) (int, byte, string) {
var remaining string var remaining string
if i := strings.IndexByte(s, ';'); i >= 0 { i := -1
if delimiter == 0 {
// Faster than strings.IndexAny(";:")
i = strings.IndexByte(s, ';')
if i < 0 {
i = strings.IndexByte(s, ':')
}
} else {
i = strings.IndexByte(s, delimiter)
}
if i >= 0 {
delimiter = s[i]
remaining = s[i+1:] remaining = s[i+1:]
s = s[:i] s = s[:i]
} }
@ -293,14 +304,14 @@ func parseAnsiCode(s string) (int, string) {
for _, ch := range []byte(s) { for _, ch := range []byte(s) {
ch -= '0' ch -= '0'
if ch > 9 { if ch > 9 {
return -1, remaining return -1, delimiter, remaining
} }
code = code*10 + int(ch) code = code*10 + int(ch)
} }
return code, remaining return code, delimiter, remaining
} }
return -1, remaining return -1, delimiter, remaining
} }
func interpretCode(ansiCode string, prevState *ansiState) ansiState { func interpretCode(ansiCode string, prevState *ansiState) ansiState {
@ -328,9 +339,10 @@ func interpretCode(ansiCode string, prevState *ansiState) ansiState {
state256 := 0 state256 := 0
ptr := &state.fg ptr := &state.fg
var delimiter byte = 0
for len(ansiCode) != 0 { for len(ansiCode) != 0 {
var num int var num int
if num, ansiCode = parseAnsiCode(ansiCode); num != -1 { if num, delimiter, ansiCode = parseAnsiCode(ansiCode, delimiter); num != -1 {
switch state256 { switch state256 {
case 0: case 0:
switch num { switch num {

View File

@ -358,6 +358,7 @@ func TestAnsiCodeStringConversion(t *testing.T) {
assert("\x1b[31m", &ansiState{fg: 4, bg: 4, lbg: -1}, "\x1b[31;44m") assert("\x1b[31m", &ansiState{fg: 4, bg: 4, lbg: -1}, "\x1b[31;44m")
assert("\x1b[1;2;31m", &ansiState{fg: 2, bg: -1, attr: tui.Reverse, lbg: -1}, "\x1b[1;2;7;31;49m") assert("\x1b[1;2;31m", &ansiState{fg: 2, bg: -1, attr: tui.Reverse, lbg: -1}, "\x1b[1;2;7;31;49m")
assert("\x1b[38;5;100;48;5;200m", nil, "\x1b[38;5;100;48;5;200m") assert("\x1b[38;5;100;48;5;200m", nil, "\x1b[38;5;100;48;5;200m")
assert("\x1b[38:5:100:48:5:200m", nil, "\x1b[38;5;100;48;5;200m")
assert("\x1b[48;5;100;38;5;200m", nil, "\x1b[38;5;200;48;5;100m") assert("\x1b[48;5;100;38;5;200m", nil, "\x1b[38;5;200;48;5;100m")
assert("\x1b[48;5;100;38;2;10;20;30;1m", nil, "\x1b[1;38;2;10;20;30;48;5;100m") assert("\x1b[48;5;100;38;2;10;20;30;1m", nil, "\x1b[1;38;2;10;20;30;48;5;100m")
assert("\x1b[48;5;100;38;2;10;20;30;7m", assert("\x1b[48;5;100;38;2;10;20;30;7m",
@ -377,7 +378,7 @@ func TestParseAnsiCode(t *testing.T) {
{"-2", "", -1}, {"-2", "", -1},
} }
for _, x := range tests { for _, x := range tests {
n, s := parseAnsiCode(x.In) n, _, s := parseAnsiCode(x.In, 0)
if n != x.N || s != x.Exp { if n != x.N || s != x.Exp {
t.Fatalf("%q: got: (%d %q) want: (%d %q)", x.In, n, s, x.N, x.Exp) t.Fatalf("%q: got: (%d %q) want: (%d %q)", x.In, n, s, x.N, x.Exp)
} }
@ -385,9 +386,9 @@ func TestParseAnsiCode(t *testing.T) {
} }
// kernel/bpf/preload/iterators/README // kernel/bpf/preload/iterators/README
const ansiBenchmarkString = "\x1b[38;5;81m\x1b[01;31m\x1b[Kkernel/\x1b[0m\x1b[38;5;81mbpf/" + const ansiBenchmarkString = "\x1b[38;5;81m\x1b[01;31m\x1b[Kkernel/\x1b[0m\x1b[38:5:81mbpf/" +
"\x1b[0m\x1b[38;5;81mpreload/\x1b[0m\x1b[38;5;81miterators/" + "\x1b[0m\x1b[38:5:81mpreload/\x1b[0m\x1b[38;5;81miterators/" +
"\x1b[0m\x1b[38;5;149mMakefile\x1b[m\x1b[K\x1b[0m" "\x1b[0m\x1b[38:5:149mMakefile\x1b[m\x1b[K\x1b[0m"
func BenchmarkNextAnsiEscapeSequence(b *testing.B) { func BenchmarkNextAnsiEscapeSequence(b *testing.B) {
b.SetBytes(int64(len(ansiBenchmarkString))) b.SetBytes(int64(len(ansiBenchmarkString)))