From 011c83956b5ffb6f0f1fd29d81ebb56cfdee6a55 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Wed, 28 Dec 2022 19:12:36 +0800 Subject: [PATCH] refactor: do some internal refactor. --- include/util/tcompare.h | 29 ++-- source/libs/scalar/src/filter.c | 18 ++- source/libs/scalar/src/sclvector.c | 4 +- source/util/src/tcompare.c | 237 ++++++++++++++++++----------- source/util/test/utilTests.cpp | 235 ++++++++++++++++++++++++++++ 5 files changed, 410 insertions(+), 113 deletions(-) diff --git a/include/util/tcompare.h b/include/util/tcompare.h index c7a3ca20f2..f92e1c3970 100644 --- a/include/util/tcompare.h +++ b/include/util/tcompare.h @@ -36,17 +36,18 @@ extern "C" { #define FLT_GREATEREQUAL(_x, _y) (FLT_EQUAL((_x), (_y)) || ((_x) > (_y))) #define FLT_LESSEQUAL(_x, _y) (FLT_EQUAL((_x), (_y)) || ((_x) < (_y))) -#define PATTERN_COMPARE_INFO_INITIALIZER \ - { '%', '_' } +#define PATTERN_COMPARE_INFO_INITIALIZER { '%', '_', L'%', L'_' } typedef struct SPatternCompareInfo { - char matchAll; // symbol for match all wildcard, default: '%' - char matchOne; // symbol for match one wildcard, default: '_' + char matchAll; // symbol for match all wildcard, default: '%' + char matchOne; // symbol for match one wildcard, default: '_' + TdUcs4 umatchAll; // unicode version matchAll + TdUcs4 umatchOne; // unicode version matchOne } SPatternCompareInfo; -int32_t patternMatch(const char *pattern, const char *str, size_t size, const SPatternCompareInfo *pInfo); +int32_t patternMatch(const char *pattern, size_t psize, const char *str, size_t ssize, const SPatternCompareInfo *pInfo); -int32_t WCSPatternMatch(const TdUcs4 *pattern, const TdUcs4 *str, size_t size, const SPatternCompareInfo *pInfo); +int32_t wcsPatternMatch(const TdUcs4 *pattern, size_t psize, const TdUcs4 *str, size_t ssize, const SPatternCompareInfo *pInfo); int32_t taosArrayCompareString(const void *a, const void *b); @@ -79,9 +80,11 @@ int32_t compareDoubleVal(const void *pLeft, const void *pRight); int32_t compareLenPrefixedStr(const void *pLeft, const void *pRight); int32_t compareLenPrefixedWStr(const void *pLeft, const void *pRight); -int32_t compareStrRegexComp(const void *pLeft, const void *pRight); -int32_t compareStrRegexCompMatch(const void *pLeft, const void *pRight); -int32_t compareStrRegexCompNMatch(const void *pLeft, const void *pRight); +int32_t comparestrRegexMatch(const void *pLeft, const void *pRight); +int32_t comparestrRegexNMatch(const void *pLeft, const void *pRight); + +int32_t comparewcsRegexMatch(const void *pLeft, const void *pRight); +int32_t comparewcsRegexNMatch(const void *pLeft, const void *pRight); int32_t compareInt8ValDesc(const void *pLeft, const void *pRight); int32_t compareInt16ValDesc(const void *pLeft, const void *pRight); @@ -99,11 +102,11 @@ int32_t compareUint64ValDesc(const void *pLeft, const void *pRight); int32_t compareLenPrefixedStrDesc(const void *pLeft, const void *pRight); int32_t compareLenPrefixedWStrDesc(const void *pLeft, const void *pRight); -int32_t compareStrPatternMatch(const void *pLeft, const void *pRight); -int32_t compareStrPatternNotMatch(const void *pLeft, const void *pRight); +int32_t comparestrPatternMatch(const void *pLeft, const void *pRight); +int32_t comparestrPatternNMatch(const void *pLeft, const void *pRight); -int32_t compareWStrPatternMatch(const void *pLeft, const void *pRight); -int32_t compareWStrPatternNotMatch(const void *pLeft, const void *pRight); +int32_t comparewcsPatternMatch(const void *pLeft, const void *pRight); +int32_t comparewcsPatternNMatch(const void *pLeft, const void *pRight); int32_t compareInt8Int16(const void *pLeft, const void *pRight); int32_t compareInt8Int32(const void *pLeft, const void *pRight); diff --git a/source/libs/scalar/src/filter.c b/source/libs/scalar/src/filter.c index 063699e8e4..8b7651765b 100644 --- a/source/libs/scalar/src/filter.c +++ b/source/libs/scalar/src/filter.c @@ -130,9 +130,9 @@ __compar_fn_t gDataCompare[] = {compareInt32Val, compareFloatVal, compareDoubleVal, compareLenPrefixedStr, - compareStrPatternMatch, + comparestrPatternMatch, compareChkInString, - compareWStrPatternMatch, + comparewcsPatternMatch, compareLenPrefixedWStr, compareUint8Val, compareUint16Val, @@ -142,15 +142,17 @@ __compar_fn_t gDataCompare[] = {compareInt32Val, setChkInBytes2, setChkInBytes4, setChkInBytes8, - compareStrRegexCompMatch, - compareStrRegexCompNMatch, + comparestrRegexMatch, + comparestrRegexNMatch, setChkNotInBytes1, setChkNotInBytes2, setChkNotInBytes4, setChkNotInBytes8, compareChkNotInString, - compareStrPatternNotMatch, - compareWStrPatternNotMatch}; + comparestrPatternNMatch, + comparewcsPatternNMatch, + comparewcsRegexMatch, + comparewcsRegexNMatch,}; __compar_fn_t gInt8SignCompare[] = {compareInt8Val, compareInt8Int16, compareInt8Int32, compareInt8Int64, compareInt8Float, compareInt8Double}; @@ -295,9 +297,9 @@ int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) { case TSDB_DATA_TYPE_NCHAR: { if (optr == OP_TYPE_MATCH) { - comparFn = 19; + comparFn = 28; } else if (optr == OP_TYPE_NMATCH) { - comparFn = 20; + comparFn = 29; } else if (optr == OP_TYPE_LIKE) { comparFn = 9; } else if (optr == OP_TYPE_NOT_LIKE) { diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c index 2eb12c8269..24ac5be845 100644 --- a/source/libs/scalar/src/sclvector.c +++ b/source/libs/scalar/src/sclvector.c @@ -916,9 +916,11 @@ int32_t vectorGetConvertType(int32_t type1, int32_t type2) { int32_t vectorConvertSingleCol(SScalarParam *input, SScalarParam *output, int32_t type, int32_t startIndex, int32_t numOfRows) { - SDataType t = {.type = type, .bytes = tDataTypes[type].bytes}; output->numOfRows = input->numOfRows; + SDataType t = {.type = type}; + t.bytes = IS_VAR_DATA_TYPE(t.type)? input->columnData->info.bytes:tDataTypes[type].bytes; + int32_t code = sclCreateColumnInfoData(&t, input->numOfRows, output); if (code != TSDB_CODE_SUCCESS) { return TSDB_CODE_OUT_OF_MEMORY; diff --git a/source/util/src/tcompare.c b/source/util/src/tcompare.c index 32e27f886b..21e7d9e0cd 100644 --- a/source/util/src/tcompare.c +++ b/source/util/src/tcompare.c @@ -1007,59 +1007,66 @@ int32_t compareJsonValDesc(const void *pLeft, const void *pRight) { return compa * '_': Matches one character * */ -int32_t patternMatch(const char *patterStr, const char *str, size_t size, const SPatternCompareInfo *pInfo) { +int32_t patternMatch(const char *pattern, size_t psize, const char *str, size_t ssize, const SPatternCompareInfo *pInfo) { char c, c1; int32_t i = 0; int32_t j = 0; - int32_t o = 0; - int32_t m = 0; + int32_t nMatchChar = 0; - while ((c = patterStr[i++]) != 0) { + while ((c = pattern[i++]) != 0 && (i <= psize)) { if (c == pInfo->matchAll) { /* Match "*" */ - while ((c = patterStr[i++]) == pInfo->matchAll || c == pInfo->matchOne) { + while ((c = pattern[i++]) == pInfo->matchAll || c == pInfo->matchOne) { + if (i > psize) { // overflow check + break; + } + if (c == pInfo->matchOne) { - if (j > size || str[j++] == 0) { - // empty string, return not match + if (j > ssize || str[j++] == 0) { // empty string, return not match return TSDB_PATTERN_NOWILDCARDMATCH; } else { - ++o; + ++nMatchChar; } } } - if (c == 0) { + if (c == 0 || i > psize) { return TSDB_PATTERN_MATCH; /* "*" at the end of the pattern matches */ } - char next[3] = {toupper(c), tolower(c), 0}; - m = o; + char acceptArray[3] = {toupper(c), tolower(c), 0}; + + str += nMatchChar; + int32_t remain = ssize - nMatchChar; while (1) { - size_t n = strcspn(str + m, next); - str += m + n; + size_t n = strcspn(str, acceptArray); - if (str[0] == 0 || (n >= size)) { + str += n; + remain -= n; + + if (str[0] == 0 || (remain <= 0)) { break; } - int32_t ret = patternMatch(&patterStr[i], ++str, size - n - 1, pInfo); + int32_t ret = patternMatch(&pattern[i], psize - i, ++str, --remain, pInfo); if (ret != TSDB_PATTERN_NOMATCH) { return ret; } - m = 0; } + return TSDB_PATTERN_NOWILDCARDMATCH; } c1 = str[j++]; - ++o; + ++nMatchChar; - if (j <= size) { - if (c == '\\' && patterStr[i] == '_' && c1 == '_') { + if (j <= ssize) { + if (c == '\\' && pattern[i] == '_' && c1 == '_') { i++; continue; } + if (c == c1 || tolower(c) == tolower(c1) || (c == pInfo->matchOne && c1 != 0)) { continue; } @@ -1068,39 +1075,52 @@ int32_t patternMatch(const char *patterStr, const char *str, size_t size, const return TSDB_PATTERN_NOMATCH; } - return (str[j] == 0 || j >= size) ? TSDB_PATTERN_MATCH : TSDB_PATTERN_NOMATCH; + return (str[j] == 0 || j >= ssize) ? TSDB_PATTERN_MATCH : TSDB_PATTERN_NOMATCH; } -int32_t WCSPatternMatch(const TdUcs4 *pattern, const TdUcs4 *str, size_t size, const SPatternCompareInfo *pInfo) { +int32_t wcsPatternMatch(const TdUcs4 *pattern, size_t psize, const TdUcs4 *str, size_t ssize, const SPatternCompareInfo *pInfo) { TdUcs4 c, c1; - TdUcs4 matchOne = L'_'; // "_" - TdUcs4 matchAll = L'%'; // "%" int32_t i = 0; int32_t j = 0; + int32_t nMatchChar = 0; - while ((c = pattern[i++]) != 0) { - if (c == matchAll) { /* Match "%" */ + while ((c = pattern[i++]) != 0 && (i <= psize)) { + /* Match "%" */ + if (c == pInfo->umatchAll) { + while ((c = pattern[i++]) == pInfo->umatchAll || c == pInfo->umatchOne) { + if (i > psize) { + break; + } - while ((c = pattern[i++]) == matchAll || c == matchOne) { - if (c == matchOne && (j >= size || str[j++] == 0)) { - return TSDB_PATTERN_NOWILDCARDMATCH; + if (c == pInfo->umatchOne) { + if (j >= ssize || str[j++] == 0) { + return TSDB_PATTERN_NOWILDCARDMATCH; + } else { + ++nMatchChar; + } } } - if (c == 0) { + + if (c == 0 || i > psize) { return TSDB_PATTERN_MATCH; } - TdUcs4 accept[3] = {towupper(c), towlower(c), 0}; + TdUcs4 acceptArray[3] = {towupper(c), towlower(c), 0}; + + str += nMatchChar; + int32_t remain = ssize - nMatchChar; while (1) { - size_t n = wcscspn(str, accept); + size_t n = wcscspn(str, acceptArray); str += n; - if (str[0] == 0 || (n >= size)) { + remain -= n; + + if (str[0] == 0 || (remain <= 0)) { break; } - int32_t ret = WCSPatternMatch(&pattern[i], ++str, size - n - 1, pInfo); + int32_t ret = wcsPatternMatch(&pattern[i], psize-i, ++str, --remain, pInfo); if (ret != TSDB_PATTERN_NOMATCH) { return ret; } @@ -1110,9 +1130,15 @@ int32_t WCSPatternMatch(const TdUcs4 *pattern, const TdUcs4 *str, size_t size, c } c1 = str[j++]; + nMatchChar++; - if (j <= size) { - if (c == c1 || towlower(c) == towlower(c1) || (c == matchOne && c1 != 0)) { + if (j <= ssize) { + if (c == L'\\' && pattern[i] == L'_' && c1 == L'_') { + i++; + continue; + } + + if (c == c1 || towlower(c) == towlower(c1) || (c == pInfo->umatchOne && c1 != 0)) { continue; } } @@ -1120,16 +1146,38 @@ int32_t WCSPatternMatch(const TdUcs4 *pattern, const TdUcs4 *str, size_t size, c return TSDB_PATTERN_NOMATCH; } - return (str[j] == 0 || j >= size) ? TSDB_PATTERN_MATCH : TSDB_PATTERN_NOMATCH; + return (str[j] == 0 || j >= ssize) ? TSDB_PATTERN_MATCH : TSDB_PATTERN_NOMATCH; } -int32_t compareStrRegexCompMatch(const void *pLeft, const void *pRight) { return compareStrRegexComp(pLeft, pRight); } +int32_t comparestrRegexNMatch(const void *pLeft, const void *pRight) { + return comparestrRegexMatch(pLeft, pRight) ? 0 : 1; +} + +static int32_t doExecRegexMatch(const char *pString, const char *pPattern) { + int32_t ret = 0; + regex_t regex; + char msgbuf[256] = {0}; + + int32_t cflags = REG_EXTENDED; + if ((ret = regcomp(®ex, pPattern, cflags)) != 0) { + regerror(ret, ®ex, msgbuf, tListLen(msgbuf)); + + uError("Failed to compile regex pattern %s. reason %s", pPattern, msgbuf); + regfree(®ex); + return 1; + } + + ret = regexec(®ex, pString, 0, NULL, 0); + if (ret != 0 && ret != REG_NOMATCH) { + regerror(ret, ®ex, msgbuf, sizeof(msgbuf)); + uDebug("Failed to match %s with pattern %s, reason %s", pString, pPattern, msgbuf) + } -int32_t compareStrRegexCompNMatch(const void *pLeft, const void *pRight) { - return compareStrRegexComp(pLeft, pRight) ? 0 : 1; + regfree(®ex); + return (ret == 0) ? 0 : 1; } -int32_t compareStrRegexComp(const void *pLeft, const void *pRight) { +int32_t comparestrRegexMatch(const void *pLeft, const void *pRight) { size_t sz = varDataLen(pRight); char *pattern = taosMemoryMalloc(sz + 1); memcpy(pattern, varDataVal(pRight), varDataLen(pRight)); @@ -1140,30 +1188,48 @@ int32_t compareStrRegexComp(const void *pLeft, const void *pRight) { memcpy(str, varDataVal(pLeft), sz); str[sz] = 0; - int32_t errCode = 0; - regex_t regex; - char msgbuf[256] = {0}; + int32_t ret = doExecRegexMatch(str, pattern); - int32_t cflags = REG_EXTENDED; - if ((errCode = regcomp(®ex, pattern, cflags)) != 0) { - regerror(errCode, ®ex, msgbuf, sizeof(msgbuf)); - uError("Failed to compile regex pattern %s. reason %s", pattern, msgbuf); - regfree(®ex); - taosMemoryFree(str); + taosMemoryFree(str); + taosMemoryFree(pattern); + + return (ret == 0) ? 0 : 1;; +} + +int32_t comparewcsRegexMatch(const void* pString, const void* pPattern) { + size_t len = varDataLen(pPattern); + char *pattern = taosMemoryMalloc(len + 1); + + int convertLen = taosUcs4ToMbs((TdUcs4 *)varDataVal(pPattern), len, pattern); + if (convertLen < 0) { taosMemoryFree(pattern); - return 1; + return TSDB_CODE_APP_ERROR; } - errCode = regexec(®ex, str, 0, NULL, 0); - if (errCode != 0 && errCode != REG_NOMATCH) { - regerror(errCode, ®ex, msgbuf, sizeof(msgbuf)); - uDebug("Failed to match %s with pattern %s, reason %s", str, pattern, msgbuf) + pattern[len] = 0; + + len = varDataLen(pString); + char *str = taosMemoryMalloc(len + 1); + convertLen = taosUcs4ToMbs((TdUcs4 *)varDataVal(pString), len, str); + if (convertLen < 0) { + taosMemoryFree(str); + taosMemoryFree(pattern); + + return TSDB_CODE_APP_ERROR; } - int32_t result = (errCode == 0) ? 0 : 1; - regfree(®ex); + + str[len] = 0; + + int32_t ret = doExecRegexMatch(str, pattern); + taosMemoryFree(str); taosMemoryFree(pattern); - return result; + + return (ret == 0) ? 0 : 1; +} + +int32_t comparewcsRegexNMatch(const void *pLeft, const void *pRight) { + return comparewcsRegexMatch(pLeft, pRight) ? 0 : 1; } int32_t taosArrayCompareString(const void *a, const void *b) { @@ -1173,46 +1239,35 @@ int32_t taosArrayCompareString(const void *a, const void *b) { return compareLenPrefixedStr(x, y); } -int32_t compareStrPatternMatch(const void *pLeft, const void *pRight) { - SPatternCompareInfo pInfo = {'%', '_'}; - - assert(varDataLen(pRight) <= TSDB_MAX_FIELD_LEN); - char *pattern = taosMemoryCalloc(varDataLen(pRight) + 1, sizeof(char)); - memcpy(pattern, varDataVal(pRight), varDataLen(pRight)); +int32_t comparestrPatternMatch(const void *pLeft, const void *pRight) { + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; + ASSERT(varDataLen(pRight) <= TSDB_MAX_FIELD_LEN); + size_t pLen = varDataLen(pRight); size_t sz = varDataLen(pLeft); - char *buf = taosMemoryMalloc(sz + 1); - memcpy(buf, varDataVal(pLeft), sz); - buf[sz] = 0; - int32_t ret = patternMatch(pattern, buf, sz, &pInfo); - taosMemoryFree(buf); - taosMemoryFree(pattern); + int32_t ret = patternMatch(varDataVal(pRight), pLen, varDataVal(pLeft), sz, &pInfo); return (ret == TSDB_PATTERN_MATCH) ? 0 : 1; } -int32_t compareStrPatternNotMatch(const void *pLeft, const void *pRight) { - return compareStrPatternMatch(pLeft, pRight) ? 0 : 1; +int32_t comparestrPatternNMatch(const void *pLeft, const void *pRight) { + return comparestrPatternMatch(pLeft, pRight) ? 0 : 1; } -int32_t compareWStrPatternMatch(const void *pLeft, const void *pRight) { - SPatternCompareInfo pInfo = {'%', '_'}; - - ASSERT(varDataLen(pRight) <= TSDB_MAX_FIELD_LEN * TSDB_NCHAR_SIZE); +int32_t comparewcsPatternMatch(const void *pLeft, const void *pRight) { + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; - char *pattern = taosMemoryCalloc(varDataLen(pRight) + TSDB_NCHAR_SIZE, 1); - memcpy(pattern, varDataVal(pRight), varDataLen(pRight)); - - int32_t ret = - WCSPatternMatch((TdUcs4 *)pattern, (TdUcs4 *)varDataVal(pLeft), varDataLen(pLeft) / TSDB_NCHAR_SIZE, &pInfo); - taosMemoryFree(pattern); + size_t psize = varDataLen(pRight); + int32_t ret = wcsPatternMatch((TdUcs4 *)varDataVal(pRight), psize / TSDB_NCHAR_SIZE, (TdUcs4 *)varDataVal(pLeft), + varDataLen(pLeft) / TSDB_NCHAR_SIZE, &pInfo); return (ret == TSDB_PATTERN_MATCH) ? 0 : 1; } -int32_t compareWStrPatternNotMatch(const void *pLeft, const void *pRight) { - return compareWStrPatternMatch(pLeft, pRight) ? 0 : 1; +int32_t comparewcsPatternNMatch(const void *pLeft, const void *pRight) { + return comparewcsPatternMatch(pLeft, pRight) ? 0 : 1; } + __compar_fn_t getComparFunc(int32_t type, int32_t optr) { __compar_fn_t comparFn = NULL; @@ -1285,13 +1340,13 @@ __compar_fn_t getComparFunc(int32_t type, int32_t optr) { break; case TSDB_DATA_TYPE_BINARY: { if (optr == OP_TYPE_MATCH) { - comparFn = compareStrRegexCompMatch; + comparFn = comparestrRegexMatch; } else if (optr == OP_TYPE_NMATCH) { - comparFn = compareStrRegexCompNMatch; + comparFn = comparestrRegexNMatch; } else if (optr == OP_TYPE_LIKE) { /* wildcard query using like operator */ - comparFn = compareStrPatternMatch; + comparFn = comparestrPatternMatch; } else if (optr == OP_TYPE_NOT_LIKE) { /* wildcard query using like operator */ - comparFn = compareStrPatternNotMatch; + comparFn = comparestrPatternNMatch; } else if (optr == OP_TYPE_IN) { comparFn = compareChkInString; } else if (optr == OP_TYPE_NOT_IN) { @@ -1305,13 +1360,13 @@ __compar_fn_t getComparFunc(int32_t type, int32_t optr) { case TSDB_DATA_TYPE_NCHAR: { if (optr == OP_TYPE_MATCH) { - comparFn = compareStrRegexCompMatch; + comparFn = comparewcsRegexMatch; } else if (optr == OP_TYPE_NMATCH) { - comparFn = compareStrRegexCompNMatch; + comparFn = comparewcsRegexNMatch; } else if (optr == OP_TYPE_LIKE) { - comparFn = compareWStrPatternMatch; + comparFn = comparewcsPatternMatch; } else if (optr == OP_TYPE_NOT_LIKE) { - comparFn = compareWStrPatternNotMatch; + comparFn = comparewcsPatternNMatch; } else if (optr == OP_TYPE_IN) { comparFn = compareChkInString; } else if (optr == OP_TYPE_NOT_IN) { diff --git a/source/util/test/utilTests.cpp b/source/util/test/utilTests.cpp index e69de29bb2..27496ff9b9 100644 --- a/source/util/test/utilTests.cpp +++ b/source/util/test/utilTests.cpp @@ -0,0 +1,235 @@ +#include +#include +#include + +#include "tarray.h" +#include "tcompare.h" + +namespace { +} // namespace + +TEST(utilTest, wchar_pattern_match_test) { + const TdWchar* pattern = L"%1"; + + int32_t ret = 0; + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; + + const TdWchar* str0 = L"14"; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str0), wcslen(str0), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* str1 = L"11"; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str1), wcslen(str1), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* str2 = L"41"; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str2), wcslen(str2), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern3 = L"%_"; + const TdWchar* str3 = L"88"; + ret = wcsPatternMatch(reinterpret_cast(pattern3), 2, reinterpret_cast(str3), wcslen(str3), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern4 = L"%___"; + const TdWchar* str4 = L"88"; + ret = wcsPatternMatch(reinterpret_cast(pattern4), 4, reinterpret_cast(str4), wcslen(str4), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern5 = L"%___"; + const TdWchar* str5 = L"883391"; + ret = wcsPatternMatch(reinterpret_cast(pattern5), 4, reinterpret_cast(str5), wcslen(str5), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern6 = L"%___66"; + const TdWchar* str6 = L"88339166"; + ret = wcsPatternMatch(reinterpret_cast(pattern6), 6, reinterpret_cast(str6), wcslen(str6), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern7 = L"%____66"; + const TdWchar* str7 = L"66166"; + ret = wcsPatternMatch(reinterpret_cast(pattern7), 7, reinterpret_cast(str7), wcslen(str7), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern8 = L"6%____66"; + const TdWchar* str8 = L"666166"; + ret = wcsPatternMatch(reinterpret_cast(pattern8), 8, reinterpret_cast(str8), wcslen(str8), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern9 = L"6\\__6"; + const TdWchar* str9 = L"6_66"; + ret = wcsPatternMatch(reinterpret_cast(pattern9), 6, reinterpret_cast(str9), wcslen(str9), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); +} + +TEST(utilTest, wchar_pattern_match_no_terminated) { + const TdWchar* pattern = L"%1 "; + + int32_t ret = 0; + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; + + const TdWchar* str0 = L"14 "; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str0), 2, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* str1 = L"11 "; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str1), 2, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* str2 = L"41 "; + ret = wcsPatternMatch(reinterpret_cast(pattern), 2, reinterpret_cast(str2), 2, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern3 = L"%_ "; + const TdWchar* str3 = L"88 "; + ret = wcsPatternMatch(reinterpret_cast(pattern3), 2, reinterpret_cast(str3), 2, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern4 = L"%___ "; + const TdWchar* str4 = L"88 "; + ret = wcsPatternMatch(reinterpret_cast(pattern4), 4, reinterpret_cast(str4), 2, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern5 = L"%___ "; + const TdWchar* str5 = L"883391 "; + ret = wcsPatternMatch(reinterpret_cast(pattern5), 4, reinterpret_cast(str5), 6, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern6 = L"%___66 "; + const TdWchar* str6 = L"88339166 "; + ret = wcsPatternMatch(reinterpret_cast(pattern6), 6, reinterpret_cast(str6), 8, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern7 = L"%____66 "; + const TdWchar* str7 = L"66166 "; + ret = wcsPatternMatch(reinterpret_cast(pattern7), 7, reinterpret_cast(str7), 5, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern8 = L"6%____66 "; + const TdWchar* str8 = L"666166 "; + ret = wcsPatternMatch(reinterpret_cast(pattern8), 8, reinterpret_cast(str8), 6, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const TdWchar* pattern9 = L"6\\_6 "; + const TdWchar* str9 = L"6_6 "; + ret = wcsPatternMatch(reinterpret_cast(pattern9), 4, reinterpret_cast(str9), 3, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const TdWchar* pattern10 = L"% "; + const TdWchar* str10 = L"6_6 "; + ret = wcsPatternMatch(reinterpret_cast(pattern10), 1, reinterpret_cast(str10), 3, &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); +} + +TEST(utilTest, char_pattern_match_test) { + const char* pattern = "%1"; + + int32_t ret = 0; + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; + + const char* str0 = "14"; + ret = patternMatch(pattern, 2, str0, strlen(str0), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* str1 = "11"; + ret = patternMatch(pattern, 2, str1, strlen(str1), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* str2 = "41"; + ret = patternMatch(pattern, 2, str2, strlen(str2), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern3 = "%_"; + const char* str3 = "88"; + ret = patternMatch(pattern3, 2, str3, strlen(str3), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern4 = "%___"; + const char* str4 = "88"; + ret = patternMatch(pattern4, 4, str4, strlen(str4), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern5 = "%___"; + const char* str5 = "883391"; + ret = patternMatch(pattern5, 4, str5, strlen(str5), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern6 = "%___66"; + const char* str6 = "88339166"; + ret = patternMatch(pattern6, 6, str6, strlen(str6), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern7 = "%____66"; + const char* str7 = "66166"; + ret = patternMatch(pattern7, 7, str7, strlen(str7), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern8 = "6%____66"; + const char* str8 = "666166"; + ret = patternMatch(pattern8, 8, str8, strlen(str8), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern9 = "6\\_6"; + const char* str9 = "6_6"; + ret = patternMatch(pattern9, 5, str9, strlen(str9), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); +} + +TEST(utilTest, char_pattern_match_no_terminated) { + const char* pattern = "%1 "; + + int32_t ret = 0; + SPatternCompareInfo pInfo = PATTERN_COMPARE_INFO_INITIALIZER; + + const char* str0 = "14"; + ret = patternMatch(pattern, 2, str0, strlen(str0), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* str1 = "11"; + ret = patternMatch(pattern, 2, str1, strlen(str1), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* str2 = "41"; + ret = patternMatch(pattern, 2, str2, strlen(str2), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern3 = "%_ "; + const char* str3 = "88"; + ret = patternMatch(pattern3, 2, str3, strlen(str3), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern4 = "%___ "; + const char* str4 = "88"; + ret = patternMatch(pattern4, 4, str4, strlen(str4), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern5 = "%___ "; + const char* str5 = "883391"; + ret = patternMatch(pattern5, 4, str5, strlen(str5), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern6 = "%___66 "; + const char* str6 = "88339166"; + ret = patternMatch(pattern6, 6, str6, strlen(str6), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern7 = "%____66 "; + const char* str7 = "66166"; + ret = patternMatch(pattern7, 7, str7, strlen(str7), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern8 = "6%____66 "; + const char* str8 = "666166"; + ret = patternMatch(pattern8, 8, str8, strlen(str8), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_NOWILDCARDMATCH); + + const char* pattern9 = "6\\_6 "; + const char* str9 = "6_6"; + ret = patternMatch(pattern9, 4, str9, strlen(str9), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); + + const char* pattern10 = "% "; + const char* str10 = "6_6"; + ret = patternMatch(pattern10, 1, str10, strlen(str10), &pInfo); + ASSERT_EQ(ret, TSDB_PATTERN_MATCH); +} \ No newline at end of file -- GitLab