AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-6576 Check code points for UCOL_IDENTICAL. Recompute processed pattern CEs if strength changes from < quaternary to >= quaternary or visa versa.

Browse Source 
X-SVN-Rev: 24749
This commit is contained in:
Eric Mader 2008-10-07 23:37:42 +00:00
parent c712edec79
commit 57e86d3127
2 changed files with 97 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
icu4c/source/i18n/usearch.cpp
View File

@ -1073,7 +1073,6 @@ inline UBool isOutOfBounds(int32_t textlength, int32_t offset)
return offset < 0 || offset > textlength;
}
#if BOYER_MOORE
/**
* Checks for identical match
* @param strsrch string search data
@ -1135,6 +1134,7 @@ inline UBool checkIdentical(const UStringSearch *strsrch, int32_t start,
return U_SUCCESS(status) && result;
}
#if BOYER_MOORE
/**
* Checks to see if the match is repeated
* @param strsrch string search data
@ -3407,12 +3407,20 @@ U_CAPI void U_EXPORT2 usearch_reset(UStringSearch *strsrch)
UBool shift;
uint32_t varTop;
// **** hack to deal w/ how processed CEs encode quaternary ****
UCollationStrength newStrength = ucol_getStrength(strsrch->collator);
if ((strsrch->strength < UCOL_QUATERNARY && newStrength >= UCOL_QUATERNARY) ||
(strsrch->strength >= UCOL_QUATERNARY && newStrength < UCOL_QUATERNARY)) {
sameCollAttribute = FALSE;
}
strsrch->strength = ucol_getStrength(strsrch->collator);
ceMask = getMask(strsrch->strength);
if (strsrch->ceMask != ceMask) {
strsrch->ceMask = ceMask;
sameCollAttribute = FALSE;
}
// if status is a failure, ucol_getAttribute returns UCOL_DEFAULT
shift = ucol_getAttribute(strsrch->collator, UCOL_ALTERNATE_HANDLING,
&status) == UCOL_SHIFTED;
@ -3887,6 +3895,10 @@ U_CAPI UBool U_EXPORT2 usearch_search(UStringSearch *strsrch,
found = FALSE;
}
if (!checkIdentical(strsrch, mStart, mLimit)) {
found = FALSE;
}
if (found) {
break;
}

101
icu4c/source/test/cintltst/usrchtst.c
View File

@ -2279,29 +2279,41 @@ static void TestSearchForNull(void)
int expectedNum;
int count = 0;
const UChar zerodigit = 0x0030; /* 0 */
const UChar nulldigit = 0x0000; /* null */
const UChar nulldigit = 0x0001; /* null */
/* static const UChar var[(length)+1]=U_DECLARE_UTF16(cs) */
const int patternlen = 1;
const int textlen = 10;
U_STRING_DECL (pattern, "0", patternlen);
U_STRING_DECL (text, "_0IS 0 OK?", textlen);
#define PATTERN_LEN 4
#define TEXT_LEN 10
U_STRING_DECL (_pattern, "IS 0", PATTERN_LEN);
U_STRING_DECL (_text, "_0IS 0 OK?", TEXT_LEN);
UChar pattern[PATTERN_LEN + 1], text[TEXT_LEN + 1];
U_STRING_INIT (pattern, "0", patternlen);
U_STRING_INIT (text, "_0IS 0 OK?", textlen);
expectedPos = 1;
expectedLen = 1;
expectedNum = 2;
U_STRING_INIT (_pattern, "IS 0", PATTERN_LEN);
U_STRING_INIT (_text, "_0IS 0 OK?", TEXT_LEN);
expectedPos = 2;
expectedLen = 4;
expectedNum = 1;
for(pos=0;pos<patternlen;pos++) {
if(pattern[pos]==zerodigit) pattern[pos] = nulldigit;
for(pos = 0; pos < PATTERN_LEN; pos++) {
if(_pattern[pos] == zerodigit) {
pattern[pos] = nulldigit;
} else {
pattern[pos] = _pattern[pos];
}
}
for(pos=0;pos<textlen;pos++) {
if(text[pos]==zerodigit) text[pos] = nulldigit;
pattern[PATTERN_LEN] = 0x0000;
for(pos = 0; pos < TEXT_LEN; pos++) {
if(_text[pos] == zerodigit) {
text[pos] = nulldigit;
} else {
text[pos] = _text[pos];
}
}
text[TEXT_LEN] = 0x0000;
ec=U_ZERO_ERROR;
ec = U_ZERO_ERROR;
/* create a US-English collator */
coll = ucol_open ("en_US", &ec);
@ -2312,7 +2324,7 @@ static void TestSearchForNull(void)
ucol_setStrength( coll, UCOL_IDENTICAL);
/* open a search looking for 0 */
search = usearch_openFromCollator (pattern, 1, text, 9, coll, NULL, &ec);
search = usearch_openFromCollator (pattern, PATTERN_LEN, text, TEXT_LEN, coll, NULL, &ec);
TEST_ASSERT (U_SUCCESS (ec));
pos = usearch_first(search, &ec);
@ -2342,6 +2354,60 @@ static void TestSearchForNull(void)
#endif
}
static void TestStrengthIdentical(void)
{
UCollator *coll;
UErrorCode ec = U_ZERO_ERROR;
UStringSearch *search;
UChar pattern[] = {0x05E9, 0x0591, 0x05E9};
UChar text[] = {0x05E9, 0x0592, 0x05E9};
int32_t pLen = sizeof (pattern) / sizeof(pattern[0]);
int32_t tLen = sizeof(text) / sizeof (text[0]);
int32_t expectedPos = 0;
int32_t expectedLen = 3;
int32_t pos;
int32_t len;
/* create a US-English collator */
coll = ucol_open ("en_US", &ec);
/* make sure we didn't fail. */
TEST_ASSERT (U_SUCCESS (ec));
ucol_setStrength( coll, UCOL_TERTIARY);
/* open a search looking for 0 */
search = usearch_openFromCollator (pattern, pLen, text, tLen, coll, NULL, &ec);
TEST_ASSERT (U_SUCCESS (ec));
pos = usearch_first(search, &ec);
len = usearch_getMatchedLength(search);
if(pos != expectedPos) {
log_err("Expected search result: %d; Got instead: %d\n", expectedPos, pos);
}
if(len != expectedLen) {
log_err("Expected search result length: %d; Got instead: %d\n", expectedLen, len);
}
/* Now try it at strength == UCOL_IDENTICAL */
ucol_setStrength(coll, UCOL_IDENTICAL);
usearch_reset(search);
pos = usearch_first(search, &ec);
len = usearch_getMatchedLength(search);
if(pos != -1) {
log_err("Expected failure for strentgh = UCOL_IDENTICAL: got %d instead.\n", pos);
}
usearch_close(search);
ucol_close(coll);
}
void addSearchTest(TestNode** root)
{
@ -2396,6 +2462,7 @@ void addSearchTest(TestNode** root)
addTest(root, &TestDiacriticMatch, "tscoll/usrchtst/TestDiacriticMatch");
addTest(root, &TestForwardBackward, "tscoll/usrchtst/TestForwardBackward");
addTest(root, &TestSearchForNull, "tscoll/usrchtst/TestSearchForNull");
addTest(root, &TestStrengthIdentical, "tscoll/usrchtst/TestStrengthIdentical");
}
#endif /* #if !UCONFIG_NO_COLLATION */