static UConverterSelector *
serializeAndUnserialize(UConverterSelector *sel, char **buffer, UErrorCode *status) {
char *new_buffer;
int32_t ser_len, ser_len2;
/* preflight */
ser_len = ucnvsel_serialize(sel, NULL, 0, status);
if (*status != U_BUFFER_OVERFLOW_ERROR) {
log_err("ucnvsel_serialize(preflighting) failed: %s\n", u_errorName(*status));
return sel;
}
new_buffer = (char *)uprv_malloc(ser_len);
*status = U_ZERO_ERROR;
ser_len2 = ucnvsel_serialize(sel, new_buffer, ser_len, status);
if (U_FAILURE(*status) || ser_len != ser_len2) {
log_err("ucnvsel_serialize() failed: %s\n", u_errorName(*status));
uprv_free(new_buffer);
return sel;
}
ucnvsel_close(sel);
uprv_free(*buffer);
*buffer = new_buffer;
sel = ucnvsel_openFromSerialized(new_buffer, ser_len, status);
if (U_FAILURE(*status)) {
log_err("ucnvsel_openFromSerialized() failed: %s\n", u_errorName(*status));
return NULL;
}
return sel;
}
/* unserialize a selector */
U_CAPI UConverterSelector* U_EXPORT2
ucnvsel_openFromSerialized(const void* buffer, int32_t length, UErrorCode* status) {
// check if already failed
if (U_FAILURE(*status)) {
return NULL;
}
// ensure args make sense!
const uint8_t *p = (const uint8_t *)buffer;
if (length <= 0 ||
(length > 0 && (p == NULL || (U_POINTER_MASK_LSB(p, 3) != 0)))
) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
// header
if (length < 32) {
// not even enough space for a minimal header
*status = U_INDEX_OUTOFBOUNDS_ERROR;
return NULL;
}
const DataHeader *pHeader = (const DataHeader *)p;
if (!(
pHeader->dataHeader.magic1==0xda &&
pHeader->dataHeader.magic2==0x27 &&
pHeader->info.dataFormat[0] == 0x43 &&
pHeader->info.dataFormat[1] == 0x53 &&
pHeader->info.dataFormat[2] == 0x65 &&
pHeader->info.dataFormat[3] == 0x6c
)) {
/* header not valid or dataFormat not recognized */
*status = U_INVALID_FORMAT_ERROR;
return NULL;
}
if (pHeader->info.formatVersion[0] != 1) {
*status = U_UNSUPPORTED_ERROR;
return NULL;
}
uint8_t* swapped = NULL;
if (pHeader->info.isBigEndian != U_IS_BIG_ENDIAN ||
pHeader->info.charsetFamily != U_CHARSET_FAMILY
) {
// swap the data
UDataSwapper *ds =
udata_openSwapperForInputData(p, length, U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, status);
int32_t totalSize = ucnvsel_swap(ds, p, -1, NULL, status);
if (U_FAILURE(*status)) {
udata_closeSwapper(ds);
return NULL;
}
if (length < totalSize) {
udata_closeSwapper(ds);
*status = U_INDEX_OUTOFBOUNDS_ERROR;
return NULL;
}
swapped = (uint8_t*)uprv_malloc(totalSize);
if (swapped == NULL) {
udata_closeSwapper(ds);
*status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
ucnvsel_swap(ds, p, length, swapped, status);
udata_closeSwapper(ds);
if (U_FAILURE(*status)) {
uprv_free(swapped);
return NULL;
}
p = swapped;
pHeader = (const DataHeader *)p;
}
if (length < (pHeader->dataHeader.headerSize + 16 * 4)) {
// not even enough space for the header and the indexes
uprv_free(swapped);
*status = U_INDEX_OUTOFBOUNDS_ERROR;
return NULL;
}
p += pHeader->dataHeader.headerSize;
length -= pHeader->dataHeader.headerSize;
// indexes
const int32_t *indexes = (const int32_t *)p;
if (length < indexes[UCNVSEL_INDEX_SIZE]) {
uprv_free(swapped);
*status = U_INDEX_OUTOFBOUNDS_ERROR;
return NULL;
}
p += UCNVSEL_INDEX_COUNT * 4;
// create and populate the selector object
UConverterSelector* sel = (UConverterSelector*)uprv_malloc(sizeof(UConverterSelector));
char **encodings =
(char **)uprv_malloc(
indexes[UCNVSEL_INDEX_NAMES_COUNT] * sizeof(char *));
if (sel == NULL || encodings == NULL) {
uprv_free(swapped);
uprv_free(sel);
uprv_free(encodings);
*status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
uprv_memset(sel, 0, sizeof(UConverterSelector));
sel->pvCount = indexes[UCNVSEL_INDEX_PV_COUNT];
sel->encodings = encodings;
sel->encodingsCount = indexes[UCNVSEL_INDEX_NAMES_COUNT];
sel->encodingStrLength = indexes[UCNVSEL_INDEX_NAMES_LENGTH];
sel->swapped = swapped;
// trie
sel->trie = utrie2_openFromSerialized(UTRIE2_16_VALUE_BITS,
p, indexes[UCNVSEL_INDEX_TRIE_SIZE], NULL,
status);
p += indexes[UCNVSEL_INDEX_TRIE_SIZE];
if (U_FAILURE(*status)) {
ucnvsel_close(sel);
return NULL;
}
// bit vectors
sel->pv = (uint32_t *)p;
p += sel->pvCount * 4;
// encoding names
char* s = (char*)p;
for (int32_t i = 0; i < sel->encodingsCount; ++i) {
sel->encodings[i] = s;
s += uprv_strlen(s) + 1;
}
p += sel->encodingStrLength;
return sel;
}
static void TestSelector()
{
TestText text;
USet* excluded_sets[3] = { NULL };
int32_t i, testCaseIdx;
if (!getAvailableNames()) {
return;
}
if (!text_open(&text)) {
releaseAvailableNames();;
}
excluded_sets[0] = uset_openEmpty();
for(i = 1 ; i < 3 ; i++) {
excluded_sets[i] = uset_open(i*30, i*30+500);
}
for(testCaseIdx = 0; testCaseIdx < UPRV_LENGTHOF(getEncodingsFns); testCaseIdx++)
{
int32_t excluded_set_id;
int32_t num_encodings;
const char **encodings = getEncodingsFns[testCaseIdx](&num_encodings);
if (getTestOption(QUICK_OPTION) && num_encodings > 25) {
uprv_free((void *)encodings);
continue;
}
/*
* for(excluded_set_id = 0 ; excluded_set_id < 3 ; excluded_set_id++)
*
* This loop was replaced by the following statement because
* the loop made the test run longer without adding to the code coverage.
* The handling of the exclusion set is independent of the
* set of encodings, so there is no need to test every combination.
*/
excluded_set_id = testCaseIdx % UPRV_LENGTHOF(excluded_sets);
{
UConverterSelector *sel_rt, *sel_fb;
char *buffer_fb = NULL;
UErrorCode status = U_ZERO_ERROR;
sel_rt = ucnvsel_open(encodings, num_encodings,
excluded_sets[excluded_set_id],
UCNV_ROUNDTRIP_SET, &status);
if (num_encodings == gCountAvailable) {
/* test the special "all converters" parameter values */
sel_fb = ucnvsel_open(NULL, 0,
excluded_sets[excluded_set_id],
UCNV_ROUNDTRIP_AND_FALLBACK_SET, &status);
} else if (uset_isEmpty(excluded_sets[excluded_set_id])) {
/* test that a NULL set gives the same results as an empty set */
sel_fb = ucnvsel_open(encodings, num_encodings,
NULL,
UCNV_ROUNDTRIP_AND_FALLBACK_SET, &status);
} else {
sel_fb = ucnvsel_open(encodings, num_encodings,
excluded_sets[excluded_set_id],
UCNV_ROUNDTRIP_AND_FALLBACK_SET, &status);
}
if (U_FAILURE(status)) {
log_err("ucnv_sel_open(encodings %ld) failed - %s\n", testCaseIdx, u_errorName(status));
ucnvsel_close(sel_rt);
uprv_free((void *)encodings);
continue;
}
text_reset(&text);
for (;;) {
UBool *manual_rt, *manual_fb;
static UChar utf16[10000];
char *s;
int32_t length8, length16;
s = text_nextString(&text, &length8);
if (s == NULL || (getTestOption(QUICK_OPTION) && text.number > 3)) {
break;
}
manual_rt = getResultsManually(encodings, num_encodings,
s, length8,
excluded_sets[excluded_set_id],
UCNV_ROUNDTRIP_SET);
manual_fb = getResultsManually(encodings, num_encodings,
s, length8,
excluded_sets[excluded_set_id],
UCNV_ROUNDTRIP_AND_FALLBACK_SET);
/* UTF-8 with length */
status = U_ZERO_ERROR;
verifyResult(ucnvsel_selectForUTF8(sel_rt, s, length8, &status), manual_rt);
verifyResult(ucnvsel_selectForUTF8(sel_fb, s, length8, &status), manual_fb);
/* UTF-8 NUL-terminated */
verifyResult(ucnvsel_selectForUTF8(sel_rt, s, -1, &status), manual_rt);
verifyResult(ucnvsel_selectForUTF8(sel_fb, s, -1, &status), manual_fb);
u_strFromUTF8(utf16, UPRV_LENGTHOF(utf16), &length16, s, length8, &status);
if (U_FAILURE(status)) {
log_err("error converting the test text (string %ld) to UTF-16 - %s\n",
(long)text.number, u_errorName(status));
} else {
if (text.number == 0) {
sel_fb = serializeAndUnserialize(sel_fb, &buffer_fb, &status);
}
if (U_SUCCESS(status)) {
/* UTF-16 with length */
verifyResult(ucnvsel_selectForString(sel_rt, utf16, length16, &status), manual_rt);
verifyResult(ucnvsel_selectForString(sel_fb, utf16, length16, &status), manual_fb);
/* UTF-16 NUL-terminated */
verifyResult(ucnvsel_selectForString(sel_rt, utf16, -1, &status), manual_rt);
verifyResult(ucnvsel_selectForString(sel_fb, utf16, -1, &status), manual_fb);
}
}
uprv_free(manual_rt);
uprv_free(manual_fb);
}
ucnvsel_close(sel_rt);
ucnvsel_close(sel_fb);
uprv_free(buffer_fb);
}
uprv_free((void *)encodings);
}
releaseAvailableNames();
text_close(&text);
for(i = 0 ; i < 3 ; i++) {
uset_close(excluded_sets[i]);
}
}
