static void TestGetRulesAndSourceSet() {
UErrorCode status = U_ZERO_ERROR;
UTransliterator *utrans = utrans_openU(transSimpleID, -1, UTRANS_FORWARD, NULL, 0, NULL, &status);
if ( U_SUCCESS(status) ) {
USet* uset;
UChar ubuf[kUBufMax];
int32_t ulen;
status = U_ZERO_ERROR;
ulen = utrans_toRules(utrans, FALSE, ubuf, kUBufMax, &status);
if ( U_FAILURE(status) || ulen <= 50 || ulen >= 100) {
log_err("FAIL: utrans_toRules unescaped, expected noErr and len 50-100, got error=%s and len=%d\n",
u_errorName(status), ulen);
}
status = U_ZERO_ERROR;
ulen = utrans_toRules(utrans, FALSE, NULL, 0, &status);
if ( status != U_BUFFER_OVERFLOW_ERROR || ulen <= 50 || ulen >= 100) {
log_err("FAIL: utrans_toRules unescaped, expected U_BUFFER_OVERFLOW_ERROR and len 50-100, got error=%s and len=%d\n",
u_errorName(status), ulen);
}
status = U_ZERO_ERROR;
ulen = utrans_toRules(utrans, TRUE, ubuf, kUBufMax, &status);
if ( U_FAILURE(status) || ulen <= 100 || ulen >= 200) {
log_err("FAIL: utrans_toRules escaped, expected noErr and len 100-200, got error=%s and len=%d\n",
u_errorName(status), ulen);
}
status = U_ZERO_ERROR;
uset = utrans_getSourceSet(utrans, FALSE, NULL, &status);
ulen = uset_toPattern(uset, ubuf, kUBufMax, FALSE, &status);
uset_close(uset);
if ( U_FAILURE(status) || ulen <= 4 || ulen >= 20) {
log_err("FAIL: utrans_getSourceSet useFilter, expected noErr and len 4-20, got error=%s and len=%d\n",
u_errorName(status), ulen);
}
status = U_ZERO_ERROR;
uset = utrans_getSourceSet(utrans, TRUE, NULL, &status);
ulen = uset_toPattern(uset, ubuf, kUBufMax, FALSE, &status);
uset_close(uset);
if ( U_FAILURE(status) || ulen <= 4 || ulen >= 20) {
log_err("FAIL: utrans_getSourceSet ignoreFilter, expected noErr and len 4-20, got error=%s and len=%d\n",
u_errorName(status), ulen);
}
utrans_close(utrans);
} else {
log_data_err("FAIL: utrans_openRules(%s) failed, error=%s (Are you missing data?)\n",
transSimpleCName, u_errorName(status));
}
}
static void expectItems(const USet* set,
const char* items) {
const char* p = items;
UChar ustr[4096], itemStr[4096];
char buf[4096];
char *pat;
UErrorCode ec;
int32_t expectedSize = 0;
int32_t itemCount = uset_getItemCount(set);
int32_t itemIndex = 0;
UChar32 start = 1, end = 0;
int32_t itemLen = 0, length;
ec = U_ZERO_ERROR;
length = uset_toPattern(set, ustr, sizeof(ustr), TRUE, &ec);
if (U_FAILURE(ec)) {
log_err("FAIL: uset_toPattern => %s\n", u_errorName(ec));
return;
}
pat=aescstrdup(ustr, length);
if (uset_isEmpty(set) != (strlen(items)==0)) {
log_data_err("FAIL: %s should return %s from isEmpty (Are you missing data?)\n",
pat,
strlen(items)==0 ? "TRUE" : "FALSE");
}
/* Don't test patterns starting with "[^" */
if (u_strlen(ustr) > 2 && ustr[1] == 0x5e /*'^'*/) {
return;
}
while (*p) {
++expectedSize;
if (start > end || start == -1) {
/* Fetch our next item */
if (itemIndex >= itemCount) {
log_data_err("FAIL: ran out of items iterating %s (Are you missing data?)\n", pat);
return;
}
itemLen = uset_getItem(set, itemIndex, &start, &end,
itemStr, sizeof(itemStr), &ec);
if (U_FAILURE(ec) || itemLen < 0) {
log_err("FAIL: uset_getItem => %s\n", u_errorName(ec));
return;
}
if (itemLen == 0) {
log_verbose("Ok: %s item %d is %c-%c\n", pat,
itemIndex, oneUCharToChar(start),
oneUCharToChar(end));
} else {
itemStr[itemLen] = 0;
u_UCharsToChars(itemStr, buf, itemLen+1);
log_verbose("Ok: %s item %d is \"%s\"\n", pat, itemIndex, buf);
}
++itemIndex;
}
if (*p=='{') {
const char* stringStart = ++p;
int32_t stringLength = 0;
char strCopy[64];
while (*p++ != '}') {
}
stringLength = (int32_t)(p - stringStart - 1);
strncpy(strCopy, stringStart, stringLength);
strCopy[stringLength] = 0;
u_charsToUChars(stringStart, ustr, stringLength);
ustr[stringLength] = 0;
if (itemLen == 0) {
log_err("FAIL: for %s expect \"%s\" next, but got a char\n",
pat, strCopy);
return;
}
if (u_strcmp(ustr, itemStr) != 0) {
log_err("FAIL: for %s expect \"%s\" next\n",
pat, strCopy);
return;
}
}
else {
UChar32 c;
u_charsToUChars(p, ustr, 1);
c = ustr[0];
if (itemLen != 0) {
log_err("FAIL: for %s expect '%c' next, but got a string\n",
pat, *p);
return;
}
if (c != start++) {
log_err("FAIL: for %s expect '%c' next\n",
pat, *p);
return;
}
++p;
}
}
if (uset_size(set) == expectedSize) {
log_verbose("Ok: %s size is %d\n", pat, expectedSize);
} else {
log_err("FAIL: %s size is %d, expected %d\n",
pat, uset_size(set), expectedSize);
}
}
static void expectContainment(const USet* set,
const char* list,
UBool isIn) {
const char* p = list;
UChar ustr[4096];
char *pat;
UErrorCode ec;
int32_t rangeStart = -1, rangeEnd = -1, length;
ec = U_ZERO_ERROR;
length = uset_toPattern(set, ustr, sizeof(ustr), TRUE, &ec);
if(U_FAILURE(ec)) {
log_err("FAIL: uset_toPattern() fails in expectContainment() - %s\n", u_errorName(ec));
return;
}
pat=aescstrdup(ustr, length);
while (*p) {
if (*p=='{') {
const char* stringStart = ++p;
int32_t stringLength = 0;
char strCopy[64];
while (*p++ != '}') {
}
stringLength = (int32_t)(p - stringStart - 1);
strncpy(strCopy, stringStart, stringLength);
strCopy[stringLength] = 0;
u_charsToUChars(stringStart, ustr, stringLength);
if (uset_containsString(set, ustr, stringLength) == isIn) {
log_verbose("Ok: %s %s \"%s\"\n", pat,
(isIn ? "contains" : "does not contain"),
strCopy);
} else {
log_data_err("FAIL: %s %s \"%s\" (Are you missing data?)\n", pat,
(isIn ? "does not contain" : "contains"),
strCopy);
}
}
else {
UChar32 c;
u_charsToUChars(p, ustr, 1);
c = ustr[0];
if (uset_contains(set, c) == isIn) {
log_verbose("Ok: %s %s '%c'\n", pat,
(isIn ? "contains" : "does not contain"),
*p);
} else {
log_data_err("FAIL: %s %s '%c' (Are you missing data?)\n", pat,
(isIn ? "does not contain" : "contains"),
*p);
}
/* Test the range API too by looking for ranges */
if (c == rangeEnd+1) {
rangeEnd = c;
} else {
if (rangeStart >= 0) {
if (uset_containsRange(set, rangeStart, rangeEnd) == isIn) {
log_verbose("Ok: %s %s U+%04X-U+%04X\n", pat,
(isIn ? "contains" : "does not contain"),
rangeStart, rangeEnd);
} else {
log_data_err("FAIL: %s %s U+%04X-U+%04X (Are you missing data?)\n", pat,
(isIn ? "does not contain" : "contains"),
rangeStart, rangeEnd);
}
}
rangeStart = rangeEnd = c;
}
++p;
}
}
if (rangeStart >= 0) {
if (uset_containsRange(set, rangeStart, rangeEnd) == isIn) {
log_verbose("Ok: %s %s U+%04X-U+%04X\n", pat,
(isIn ? "contains" : "does not contain"),
rangeStart, rangeEnd);
} else {
log_data_err("FAIL: %s %s U+%04X-U+%04X (Are you missing data?)\n", pat,
(isIn ? "does not contain" : "contains"),
rangeStart, rangeEnd);
}
}
}
