basic/utf8: add ascii_is_valid_n()

This commit is contained in:
Zbigniew Jędrzejewski-Szmek
2018-06-09 13:41:24 +02:00
parent d23c3e4c28
commit 294a3121aa
3 changed files with 31 additions and 1 deletions
+18
View File
@@ -247,6 +247,9 @@ char *utf8_escape_non_printable(const char *str) {
char *ascii_is_valid(const char *str) {
const char *p;
/* Check whether the string consists of valid ASCII bytes,
* i.e values between 0 and 127, inclusive. */
assert(str);
for (p = str; *p; p++)
@@ -256,6 +259,21 @@ char *ascii_is_valid(const char *str) {
return (char*) str;
}
char *ascii_is_valid_n(const char *str, size_t len) {
size_t i;
/* Very similar to ascii_is_valid(), but checks exactly len
* bytes and rejects any NULs in that range. */
assert(str);
for (i = 0; i < len; i++)
if ((unsigned char) str[i] >= 128 || str[i] == 0)
return NULL;
return (char*) str;
}
/**
* utf8_encode_unichar() - Encode single UCS-4 character as UTF-8
* @out_utf8: output buffer of at least 4 bytes or NULL
+1
View File
@@ -22,6 +22,7 @@ bool unichar_is_valid(char32_t c);
const char *utf8_is_valid(const char *s) _pure_;
char *ascii_is_valid(const char *s) _pure_;
char *ascii_is_valid_n(const char *str, size_t len);
bool utf8_is_printable_newline(const char* str, size_t length, bool newline) _pure_;
#define utf8_is_printable(str, length) utf8_is_printable_newline(str, length, true)
+12 -1
View File
@@ -24,11 +24,21 @@ static void test_utf8_is_valid(void) {
}
static void test_ascii_is_valid(void) {
assert_se(ascii_is_valid("alsdjf\t\vbarr\nba z"));
assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
assert_se(!ascii_is_valid("\342\204\242"));
assert_se(!ascii_is_valid("\341\204"));
}
static void test_ascii_is_valid_n(void) {
assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
assert_se(!ascii_is_valid_n("\342\204\242", 3));
assert_se(!ascii_is_valid_n("\342\204\242", 2));
assert_se(!ascii_is_valid_n("\342\204\242", 1));
assert_se( ascii_is_valid_n("\342\204\242", 0));
}
static void test_utf8_encoded_valid_unichar(void) {
assert_se(utf8_encoded_valid_unichar("\342\204\242") == 3);
assert_se(utf8_encoded_valid_unichar("\302\256") == 2);
@@ -115,6 +125,7 @@ int main(int argc, char *argv[]) {
test_utf8_is_valid();
test_utf8_is_printable();
test_ascii_is_valid();
test_ascii_is_valid_n();
test_utf8_encoded_valid_unichar();
test_utf8_escaping();
test_utf8_escaping_printable();