virtual void run() {
//sleep(10000);
int32_t line = 0;
uint8_t sk1[1024], sk2[1024];
uint8_t *oldSk = NULL, *newSk = sk1;
int32_t resLen = 0, oldLen = 0;
int32_t i = 0;
for(i = 0; i < noLines; i++) {
resLen = ucol_getSortKey(coll, lines[i].buff, lines[i].buflen, newSk, 1024);
int32_t res = 0, cmpres = 0, cmpres2 = 0;
if(oldSk != NULL) {
res = strcmp((char *)oldSk, (char *)newSk);
cmpres = ucol_strcoll(coll, lines[i-1].buff, lines[i-1].buflen, lines[i].buff, lines[i].buflen);
cmpres2 = ucol_strcoll(coll, lines[i].buff, lines[i].buflen, lines[i-1].buff, lines[i-1].buflen);
//cmpres = res;
//cmpres2 = -cmpres;
if(cmpres != -cmpres2) {
error("Compare result not symmetrical on line "+ line);
break;
}
if(((res&0x80000000) != (cmpres&0x80000000)) || (res == 0 && cmpres != 0) || (res != 0 && cmpres == 0)) {
error(UnicodeString("Difference between ucol_strcoll and sortkey compare on line ")+ UnicodeString(line));
break;
}
if(res > 0) {
error(UnicodeString("Line %i is not greater or equal than previous line ")+ UnicodeString(i));
break;
} else if(res == 0) { /* equal */
res = u_strcmpCodePointOrder(lines[i-1].buff, lines[i].buff);
if (res == 0) {
error(UnicodeString("Probable error in test file on line %i (comparing identical strings)")+ UnicodeString(i));
break;
}
/*
* UCA 6.0 test files can have lines that compare == if they are
* different strings but canonically equivalent.
else if (res > 0) {
error(UnicodeString("Sortkeys are identical, but code point compare gives >0 on line ")+ UnicodeString(i));
break;
}
*/
}
}
oldSk = newSk;
oldLen = resLen;
newSk = (newSk == sk1)?sk2:sk1;
}
}
virtual void run() {
uint8_t sk1[1024], sk2[1024];
uint8_t *oldSk = NULL, *newSk = sk1;
int32_t oldLen = 0;
int32_t prev = 0;
int32_t i = 0;
for(i = 0; i < noLines; i++) {
if(lines[i].buflen == 0) { continue; }
if(skipLineBecauseOfBug(lines[i].buff, lines[i].buflen)) { continue; }
int32_t resLen = coll->getSortKey(lines[i].buff, lines[i].buflen, newSk, 1024);
if(oldSk != NULL) {
int32_t skres = strcmp((char *)oldSk, (char *)newSk);
int32_t cmpres = coll->compare(lines[prev].buff, lines[prev].buflen, lines[i].buff, lines[i].buflen);
int32_t cmpres2 = coll->compare(lines[i].buff, lines[i].buflen, lines[prev].buff, lines[prev].buflen);
if(cmpres != -cmpres2) {
IntlTest::gTest->errln(UnicodeString("Compare result not symmetrical on line ") + (i + 1));
break;
}
if(cmpres != normalizeResult(skres)) {
IntlTest::gTest->errln(UnicodeString("Difference between coll->compare and sortkey compare on line ") + (i + 1));
break;
}
int32_t res = cmpres;
if(res == 0 && !isAtLeastUCA62) {
// Up to UCA 6.1, the collation test files use a custom tie-breaker,
// comparing the raw input strings.
res = u_strcmpCodePointOrder(lines[prev].buff, lines[i].buff);
// Starting with UCA 6.2, the collation test files use the standard UCA tie-breaker,
// comparing the NFD versions of the input strings,
// which we do via setting strength=identical.
}
if(res > 0) {
IntlTest::gTest->errln(UnicodeString("Line is not greater or equal than previous line, for line ") + (i + 1));
break;
}
}
oldSk = newSk;
oldLen = resLen;
(void)oldLen; // Suppress set but not used warning.
prev = i;
newSk = (newSk == sk1)?sk2:sk1;
}
}
void UCAConformanceTest::testConformance(const Collator *coll)
{
if(testFile == 0) {
return;
}
uint32_t skipFlags = 0;
if(coll->getAttribute(UCOL_ALTERNATE_HANDLING, status) == UCOL_SHIFTED) {
skipFlags |= IS_SHIFTED;
}
if(coll == rbUCA) {
skipFlags |= FROM_RULES;
}
logln("-prop:ucaconfnosortkeys=1 turns off getSortKey() in UCAConformanceTest");
UBool withSortKeys = getProperty("ucaconfnosortkeys") == NULL;
int32_t line = 0;
UChar b1[1024], b2[1024];
UChar *buffer = b1, *oldB = NULL;
char lineB1[1024], lineB2[1024];
char *lineB = lineB1, *oldLineB = lineB2;
uint8_t sk1[1024], sk2[1024];
uint8_t *oldSk = NULL, *newSk = sk1;
int32_t oldLen = 0;
int32_t oldBlen = 0;
uint32_t first = 0;
while (fgets(lineB, 1024, testFile) != NULL) {
// remove trailing whitespace
u_rtrim(lineB);
line++;
if(*lineB == 0 || lineB[0] == '#') {
continue;
}
int32_t buflen = u_parseString(lineB, buffer, 1024, &first, &status);
if(U_FAILURE(status)) {
errln("Error parsing line %ld (%s): %s\n",
(long)line, u_errorName(status), lineB);
status = U_ZERO_ERROR;
}
buffer[buflen] = 0;
if(skipLineBecauseOfBug(buffer, buflen, skipFlags)) {
logln("Skipping line %i because of a known bug", line);
continue;
}
int32_t resLen = withSortKeys ? coll->getSortKey(buffer, buflen, newSk, 1024) : 0;
if(oldSk != NULL) {
UBool ok=TRUE;
int32_t skres = withSortKeys ? strcmp((char *)oldSk, (char *)newSk) : 0;
int32_t cmpres = coll->compare(oldB, oldBlen, buffer, buflen, status);
int32_t cmpres2 = coll->compare(buffer, buflen, oldB, oldBlen, status);
if(cmpres != -cmpres2) {
errln("Compare result not symmetrical on line %i: "
"previous vs. current (%d) / current vs. previous (%d)",
line, cmpres, cmpres2);
ok = FALSE;
}
// TODO: Compare with normalization turned off if the input passes the FCD test.
if(withSortKeys && cmpres != normalizeResult(skres)) {
errln("Difference between coll->compare (%d) and sortkey compare (%d) on line %i",
cmpres, skres, line);
ok = FALSE;
}
int32_t res = cmpres;
if(res == 0 && !isAtLeastUCA62) {
// Up to UCA 6.1, the collation test files use a custom tie-breaker,
// comparing the raw input strings.
res = u_strcmpCodePointOrder(oldB, buffer);
// Starting with UCA 6.2, the collation test files use the standard UCA tie-breaker,
// comparing the NFD versions of the input strings,
// which we do via setting strength=identical.
}
if(res > 0) {
errln("Line %i is not greater or equal than previous line", line);
ok = FALSE;
}
if(!ok) {
errln(" Previous data line %s", oldLineB);
errln(" Current data line %s", lineB);
if(withSortKeys) {
UnicodeString oldS, newS;
prettify(CollationKey(oldSk, oldLen), oldS);
prettify(CollationKey(newSk, resLen), newS);
errln(" Previous key: "+oldS);
errln(" Current key: "+newS);
}
}
}
// swap buffers
oldLineB = lineB;
oldB = buffer;
oldSk = newSk;
if(lineB == lineB1) {
lineB = lineB2;
buffer = b2;
newSk = sk2;
} else {
lineB = lineB1;
buffer = b1;
newSk = sk1;
}
oldLen = resLen;
oldBlen = buflen;
}
}
void UCAConformanceTest::testConformance(UCollator *coll)
{
if(testFile == 0) {
return;
}
int32_t line = 0;
UChar b1[1024], b2[1024];
char lineB[1024];
UChar *buffer = b1, *oldB = NULL;
uint8_t sk1[1024], sk2[1024];
uint8_t *oldSk = NULL, *newSk = sk1;
int32_t resLen = 0, oldLen = 0;
int32_t buflen = 0, oldBlen = 0;
uint32_t first = 0;
uint32_t offset = 0;
while (fgets(lineB, 1024, testFile) != NULL) {
offset = 0;
line++;
if(*lineB == 0 || lineB[0] == '#') {
continue;
}
offset = u_parseString(lineB, buffer, 1024, &first, &status);
buflen = offset;
buffer[offset++] = 0;
resLen = ucol_getSortKey(coll, buffer, buflen, newSk, 1024);
int32_t res = 0, cmpres = 0, cmpres2 = 0;
if(oldSk != NULL) {
res = strcmp((char *)oldSk, (char *)newSk);
cmpres = ucol_strcoll(coll, oldB, oldBlen, buffer, buflen);
cmpres2 = ucol_strcoll(coll, buffer, buflen, oldB, oldBlen);
if(cmpres != -cmpres2) {
errln("Compare result not symmetrical on line %i", line);
}
if(((res&0x80000000) != (cmpres&0x80000000)) || (res == 0 && cmpres != 0) || (res != 0 && cmpres == 0)) {
errln("Difference between ucol_strcoll and sortkey compare on line %i", line);
logln("Data line %s", lineB);
}
if(res > 0) {
errln("Line %i is not greater or equal than previous line", line);
logln("Data line %s", lineB);
} else if(res == 0) { /* equal */
res = u_strcmpCodePointOrder(oldB, buffer);
if (res == 0) {
errln("Probable error in test file on line %i (comparing identical strings)", line);
logln("Data line %s", lineB);
} else if (res > 0) {
errln("Sortkeys are identical, but code point comapare gives >0 on line %i", line);
logln("Data line %s", lineB);
}
}
}
oldSk = newSk;
oldLen = resLen;
newSk = (newSk == sk1)?sk2:sk1;
oldB = buffer;
oldBlen = buflen;
buffer = (buffer == b1)?b2:b1;
}
}
