Fix getTokenExclude hang on a trailing backslash

getTokenExclude looped forever when the data ended in a backslash: once
the cursor hit the end, the "previous character was a backslash" check
never changed. It also treated an escaped backslash (\\) before the
terminator as escaping it, so a token could never end in a backslash.

Stop at the end of the data, and treat a terminator as escaped only when
an odd run of backslashes precedes it. Escapes are still returned as
written, so tokens without those two cases are unchanged
(testing/zstring_test output is identical).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
brad ArantandClaude Opus 5.5 committed 2026-11-20 19:58:02 -08:00
1 parent f5aee23f5e
commit 6719b8c9a3
1 file changed
+15 -7
+15 -7
View File
@@ -328,16 +328,24 @@ namespace coreutils {
} }
ZString ZString::getTokenExclude(const char *exclude) { ZString ZString::getTokenExclude(const char *exclude) {
char previous = 0; // Scans up to the first character in exclude that is not escaped.
// A terminator is escaped only by an odd run of backslashes before
// it, so "a\\" ends at its closing quote while "a\"b" does not.
// Escapes are left in the returned token as written. Stops at the
// end of the data - previously a token ending in a backslash looped
// forever there.
char *start = cursor; char *start = cursor;
do { char *end = data + length;
while ((cursor < (data + length)) && !isCharacter(*cursor, exclude)) { while(cursor < end) {
previous = *cursor; if(isCharacter(*cursor, exclude)) {
++cursor; size_t backslashes = 0;
for(char *p = cursor; (p > start) && (p[-1] == '\\'); --p)
++backslashes;
if((backslashes % 2) == 0)
break;
} }
++cursor; ++cursor;
} while(previous == '\\'); }
--cursor;
return ZString(start, cursor - start); return ZString(start, cursor - start);
} }