string-util: also check for 0x1b 0x5c ST when stripping ANSI from strings

This commit is contained in:
Lennart Poettering
2024-10-31 11:38:18 +01:00
parent 0367424786
commit a39c51799b
2 changed files with 31 additions and 3 deletions
+22 -3
View File
@@ -699,6 +699,7 @@ char* strip_tab_ansi(char **ibuf, size_t *_isz, size_t highlight[2]) {
STATE_ESCAPE,
STATE_CSI,
STATE_OSC,
STATE_OSC_CLOSING,
} state = STATE_OTHER;
_cleanup_(memstream_done) MemStream m = {};
size_t isz, shift[2] = {}, n_carriage_returns = 0;
@@ -711,7 +712,7 @@ char* strip_tab_ansi(char **ibuf, size_t *_isz, size_t highlight[2]) {
*
* 1. Replaces TABs by 8 spaces
* 2. Strips ANSI color sequences (a subset of CSI), i.e. ESC '[' … 'm' sequences
* 3. Strips ANSI operating system sequences (OSC), i.e. ESC ']' … BEL sequences
* 3. Strips ANSI operating system sequences (OSC), i.e. ESC ']' … ST sequences
* 4. Strip trailing \r characters (since they would "move the cursor", but have no
* other effect).
*
@@ -796,14 +797,32 @@ char* strip_tab_ansi(char **ibuf, size_t *_isz, size_t highlight[2]) {
case STATE_OSC:
assert(n_carriage_returns == 0);
/* There are three kinds of OSC terminators: \x07, \x1b\x5c or \x9c. We only support
* the first two, because the last one is a valid UTF-8 codepoint and hence creates
* an ambiguity (many Terminal emulators refuse to support it as well). */
if (i >= *ibuf + isz || /* EOT … */
(*i != '\a' && (uint8_t) *i < 32U) || (uint8_t) *i > 126U) { /* … or invalid chars in sequence */
(!IN_SET(*i, '\x07', '\x1b') && (uint8_t) *i < 32U) || (uint8_t) *i > 126U) { /* … or invalid chars in sequence */
fputc('\x1B', f);
fputc(']', f);
advance_offsets(i - *ibuf, highlight, shift, 2);
state = STATE_OTHER;
i = begin-1;
} else if (*i == '\a')
} else if (*i == '\x07') /* Single character ST */
state = STATE_OTHER;
else if (*i == '\x1B')
state = STATE_OSC_CLOSING;
break;
case STATE_OSC_CLOSING:
if (i >= *ibuf + isz || /* EOT … */
*i != '\x5c') { /* … or incomplete two-byte ST in sequence */
fputc('\x1B', f);
fputc(']', f);
advance_offsets(i - *ibuf, highlight, shift, 2);
state = STATE_OTHER;
i = begin-1;
} else if (*i == '\x5c')
state = STATE_OTHER;
break;