static void
test_utf8proc_normalization(TestBatchRunner *runner) {
SKIP(runner, 1,
"utf8proc can't handle control chars or Unicode non-chars");
return;
for (int32_t i = 0; i < 100; i++) {
String *source = TestUtils_random_string(rand() % 40);
// Normalize once.
uint8_t *normalized;
int32_t check = utf8proc_map((const uint8_t*)Str_Get_Ptr8(source),
Str_Get_Size(source),
&normalized,
UTF8PROC_STABLE |
UTF8PROC_COMPOSE |
UTF8PROC_COMPAT |
UTF8PROC_CASEFOLD);
if (check < 0) {
lucy_Json_set_tolerant(1);
String *json = lucy_Json_to_json((Obj*)source);
if (!json) {
json = Str_newf("[failed to encode]");
}
FAIL(runner, "Failed to normalize: %s", Str_Get_Ptr8(json));
DECREF(json);
DECREF(source);
return;
}
// Normalize again.
size_t normalized_len = strlen((char*)normalized);
uint8_t *dupe;
int32_t dupe_check = utf8proc_map(normalized, normalized_len, &dupe,
UTF8PROC_STABLE |
UTF8PROC_COMPOSE |
UTF8PROC_COMPAT |
UTF8PROC_CASEFOLD);
if (dupe_check < 0) {
THROW(ERR, "Unexpected normalization error: %i32", dupe_check);
}
int comparison = strcmp((char*)normalized, (char*)dupe);
free(dupe);
free(normalized);
DECREF(source);
if (comparison != 0) {
FAIL(runner, "Not fully normalized");
return;
}
}
PASS(runner, "Normalization successful.");
}
static void
test_stress(TestBatch *batch) {
Hash *hash = Hash_new(0); // trigger multiple rebuilds.
VArray *expected = VA_new(1000);
VArray *keys;
VArray *values;
for (uint32_t i = 0; i < 1000; i++) {
CharBuf *cb = TestUtils_random_string(rand() % 1200);
while (Hash_Fetch(hash, (Obj*)cb)) {
DECREF(cb);
cb = TestUtils_random_string(rand() % 1200);
}
Hash_Store(hash, (Obj*)cb, (Obj*)cb);
VA_Push(expected, INCREF(cb));
}
VA_Sort(expected, NULL, NULL);
// Overwrite for good measure.
for (uint32_t i = 0; i < 1000; i++) {
CharBuf *cb = (CharBuf*)VA_Fetch(expected, i);
Hash_Store(hash, (Obj*)cb, INCREF(cb));
}
keys = Hash_Keys(hash);
values = Hash_Values(hash);
VA_Sort(keys, NULL, NULL);
VA_Sort(values, NULL, NULL);
TEST_TRUE(batch, VA_Equals(keys, (Obj*)expected), "stress Keys");
TEST_TRUE(batch, VA_Equals(values, (Obj*)expected), "stress Values");
DECREF(keys);
DECREF(values);
DECREF(expected);
DECREF(hash);
}
