basic/extract-word: add EXTRACT_UNESCAPE_SEPARATORS mode
This allows separators to be escaped, for example to allow "a\:b:c", to be treated as "a:b", "c" with ":" as the separator.
This commit is contained in:
committed by
Luca Boccassi
parent
866fdcceb4
commit
1e198efcdb
+15
-10
@@ -86,25 +86,30 @@ int extract_first_word(const char **p, char **ret, const char *separators, Extra
|
||||
return -EINVAL;
|
||||
}
|
||||
|
||||
if (flags & EXTRACT_CUNESCAPE) {
|
||||
if (flags & (EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS)) {
|
||||
bool eight_bit = false;
|
||||
char32_t u;
|
||||
|
||||
r = cunescape_one(*p, (size_t) -1, &u, &eight_bit, false);
|
||||
if (r < 0) {
|
||||
if (flags & EXTRACT_CUNESCAPE_RELAX) {
|
||||
s[sz++] = '\\';
|
||||
s[sz++] = c;
|
||||
} else
|
||||
return -EINVAL;
|
||||
} else {
|
||||
if ((flags & EXTRACT_CUNESCAPE) &&
|
||||
(r = cunescape_one(*p, (size_t) -1, &u, &eight_bit, false)) >= 0) {
|
||||
/* A valid escaped sequence */
|
||||
assert(r >= 1);
|
||||
|
||||
(*p) += r - 1;
|
||||
|
||||
if (eight_bit)
|
||||
s[sz++] = u;
|
||||
else
|
||||
sz += utf8_encode_unichar(s + sz, u);
|
||||
}
|
||||
} else if ((flags & EXTRACT_UNESCAPE_SEPARATORS) &&
|
||||
strchr(separators, **p))
|
||||
/* An escaped separator char */
|
||||
s[sz++] = c;
|
||||
else if (flags & EXTRACT_CUNESCAPE_RELAX) {
|
||||
s[sz++] = '\\';
|
||||
s[sz++] = c;
|
||||
} else
|
||||
return -EINVAL;
|
||||
} else
|
||||
s[sz++] = c;
|
||||
|
||||
|
||||
@@ -7,9 +7,10 @@ typedef enum ExtractFlags {
|
||||
EXTRACT_RELAX = 1 << 0,
|
||||
EXTRACT_CUNESCAPE = 1 << 1,
|
||||
EXTRACT_CUNESCAPE_RELAX = 1 << 2,
|
||||
EXTRACT_UNQUOTE = 1 << 3,
|
||||
EXTRACT_DONT_COALESCE_SEPARATORS = 1 << 4,
|
||||
EXTRACT_RETAIN_ESCAPE = 1 << 5,
|
||||
EXTRACT_UNESCAPE_SEPARATORS = 1 << 3,
|
||||
EXTRACT_UNQUOTE = 1 << 4,
|
||||
EXTRACT_DONT_COALESCE_SEPARATORS = 1 << 5,
|
||||
EXTRACT_RETAIN_ESCAPE = 1 << 6,
|
||||
} ExtractFlags;
|
||||
|
||||
int extract_first_word(const char **p, char **ret, const char *separators, ExtractFlags flags);
|
||||
|
||||
@@ -341,6 +341,46 @@ static void test_extract_first_word(void) {
|
||||
assert_se(streq(t, "foo\\xbar"));
|
||||
free(t);
|
||||
assert_se(p == NULL);
|
||||
|
||||
p = "\\:";
|
||||
assert_se(extract_first_word(&p, &t, ":", EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS) == 1);
|
||||
assert_se(streq(t, ":"));
|
||||
free(t);
|
||||
assert_se(p == NULL);
|
||||
|
||||
p = "a\\:b";
|
||||
assert_se(extract_first_word(&p, &t, ":", EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS) == 1);
|
||||
assert_se(streq(t, "a:b"));
|
||||
free(t);
|
||||
assert_se(p == NULL);
|
||||
|
||||
p = "a\\ b:c";
|
||||
assert_se(extract_first_word(&p, &t, WHITESPACE ":", EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS) == 1);
|
||||
assert_se(streq(t, "a b"));
|
||||
free(t);
|
||||
assert_se(extract_first_word(&p, &t, WHITESPACE ":", EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS) == 1);
|
||||
assert_se(streq(t, "c"));
|
||||
free(t);
|
||||
assert_se(p == NULL);
|
||||
|
||||
p = "\\:";
|
||||
assert_se(extract_first_word(&p, &t, ":", EXTRACT_CUNESCAPE) == -EINVAL);
|
||||
|
||||
p = "a\\:b";
|
||||
assert_se(extract_first_word(&p, &t, ":", EXTRACT_CUNESCAPE) == -EINVAL);
|
||||
assert_se(extract_first_word(&p, &t, ":", EXTRACT_CUNESCAPE) == 1);
|
||||
assert_se(streq(t, "b"));
|
||||
free(t);
|
||||
|
||||
p = "a\\ b:c";
|
||||
assert_se(extract_first_word(&p, &t, WHITESPACE ":", EXTRACT_CUNESCAPE) == -EINVAL);
|
||||
assert_se(extract_first_word(&p, &t, WHITESPACE ":", EXTRACT_CUNESCAPE) == 1);
|
||||
assert_se(streq(t, "b"));
|
||||
free(t);
|
||||
assert_se(extract_first_word(&p, &t, WHITESPACE ":", EXTRACT_CUNESCAPE) == 1);
|
||||
assert_se(streq(t, "c"));
|
||||
free(t);
|
||||
assert_se(p == NULL);
|
||||
}
|
||||
|
||||
static void test_extract_first_word_and_warn(void) {
|
||||
|
||||
Reference in New Issue
Block a user