Merge pull request #7826 from taosdata/feature/szhou/regex-nmatch

support match/nmatch relation between binary/nchar tbname/tag/col and posix regular expression

src/client/src/tscSQLParser.c

@@ -281,6 +281,8 @@ static uint8_t convertRelationalOperator(SStrToken *pToken) {
       return TSDB_RELATION_LIKE;
     case TK_MATCH:
       return TSDB_RELATION_MATCH;
+    case TK_NMATCH:
+      return TSDB_RELATION_NMATCH;
     case TK_ISNULL:
       return TSDB_RELATION_ISNULL;
     case TK_NOTNULL:
@@ -3782,6 +3784,9 @@ static int32_t doExtractColumnFilterInfo(SSqlCmd* pCmd, SQueryInfo* pQueryInfo,
     case TK_MATCH:
       pColumnFilter->lowerRelOptr = TSDB_RELATION_MATCH;
       break;
+    case TK_NMATCH:
+      pColumnFilter->lowerRelOptr = TSDB_RELATION_NMATCH;
+      break;
     case TK_ISNULL:
       pColumnFilter->lowerRelOptr = TSDB_RELATION_ISNULL;
       break;
@@ -3846,15 +3851,17 @@ static int32_t tablenameListToString(tSqlExpr* pExpr, SStringBuilder* sb) {
 }
 
 static int32_t tablenameCondToString(tSqlExpr* pExpr, uint32_t opToken, SStringBuilder* sb) {
-  assert(opToken == TK_LIKE || opToken == TK_MATCH);
+  assert(opToken == TK_LIKE || opToken == TK_MATCH || opToken == TK_NMATCH);
   if (opToken == TK_LIKE) {
     taosStringBuilderAppendStringLen(sb, QUERY_COND_REL_PREFIX_LIKE, QUERY_COND_REL_PREFIX_LIKE_LEN);
     taosStringBuilderAppendString(sb, pExpr->value.pz);
   } else if (opToken == TK_MATCH) {
     taosStringBuilderAppendStringLen(sb, QUERY_COND_REL_PREFIX_MATCH, QUERY_COND_REL_PREFIX_MATCH_LEN);
     taosStringBuilderAppendString(sb, pExpr->value.pz);
+  } else if (opToken == TK_NMATCH) {
+    taosStringBuilderAppendStringLen(sb, QUERY_COND_REL_PREFIX_NMATCH, QUERY_COND_REL_PREFIX_NMATCH_LEN);
+    taosStringBuilderAppendString(sb, pExpr->value.pz);
   }
-
   return TSDB_CODE_SUCCESS;
 }
 
@@ -3903,12 +3910,13 @@ static int32_t checkColumnFilterInfo(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, SCol
       && pExpr->tokenId != TK_NOTNULL
       && pExpr->tokenId != TK_LIKE
       && pExpr->tokenId != TK_MATCH
+      && pExpr->tokenId != TK_NMATCH
       && pExpr->tokenId != TK_IN) {
       ret = invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg2);
       goto _err_ret;
     }
   } else {
-    if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH) {
+    if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH || pExpr->tokenId == TK_NMATCH) {
       ret = invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg1);
       goto _err_ret;
     }
@@ -3956,7 +3964,7 @@ static int32_t getTablenameCond(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, tSqlExpr*
 
   if (pTableCond->tokenId == TK_IN) {
     ret = tablenameListToString(pRight, sb);
-  } else if (pTableCond->tokenId == TK_LIKE || pTableCond->tokenId == TK_MATCH) {
+  } else if (pTableCond->tokenId == TK_LIKE || pTableCond->tokenId == TK_MATCH || pTableCond->tokenId == TK_NMATCH) {
     if (pRight->tokenId != TK_STRING) {
       return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg1);
     }
@@ -4410,7 +4418,7 @@ static bool validateJoinExprNode(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, tSqlExpr
 }
 
 static bool validTableNameOptr(tSqlExpr* pExpr) {
-  const char nameFilterOptr[] = {TK_IN, TK_LIKE, TK_MATCH};
+  const char nameFilterOptr[] = {TK_IN, TK_LIKE, TK_MATCH, TK_NMATCH};
 
   for (int32_t i = 0; i < tListLen(nameFilterOptr); ++i) {
     if (pExpr->tokenId == nameFilterOptr[i]) {
@@ -4505,13 +4513,13 @@ static int32_t validateLikeExpr(tSqlExpr* pExpr, STableMeta* pTableMeta, int32_t
 // check for match expression
 static int32_t validateMatchExpr(tSqlExpr* pExpr, STableMeta* pTableMeta, int32_t index, char* msgBuf) {
   const char* msg1 = "regular expression string should be less than %d characters";
-  const char* msg2 = "illegal column type for match";
+  const char* msg2 = "illegal column type for match/nmatch";
   const char* msg3 = "invalid regular expression";
 
   tSqlExpr* pLeft  = pExpr->pLeft;
   tSqlExpr* pRight = pExpr->pRight;
 
-  if (pExpr->tokenId == TK_MATCH) {
+  if (pExpr->tokenId == TK_MATCH || pExpr->tokenId == TK_NMATCH) {
     if (pRight->value.nLen > tsMaxRegexStringLen) {
       char tmp[64] = {0};
       sprintf(tmp, msg1, tsMaxRegexStringLen);
@@ -4519,10 +4527,14 @@ static int32_t validateMatchExpr(tSqlExpr* pExpr, STableMeta* pTableMeta, int32_
     }
 
     SSchema* pSchema = tscGetTableSchema(pTableMeta);
-    if ((!isTablenameToken(&pLeft->columnName)) && !IS_VAR_DATA_TYPE(pSchema[index].type)) {
+    if ((!isTablenameToken(&pLeft->columnName)) &&(pSchema[index].type != TSDB_DATA_TYPE_BINARY)) {
       return invalidOperationMsg(msgBuf, msg2);
     }
 
+    if (!(pRight->type == SQL_NODE_VALUE && pRight->value.nType == TSDB_DATA_TYPE_BINARY)) {
+      return invalidOperationMsg(msgBuf, msg3);
+    }
+
     int errCode = 0;
     regex_t regex;
     char    regErrBuf[256] = {0};
@@ -4925,9 +4937,12 @@ static int32_t setTableCondForSTableQuery(SSqlCmd* pCmd, SQueryInfo* pQueryInfo,
   STagCond* pTagCond = &pQueryInfo->tagCond;
   pTagCond->tbnameCond.uid = pTableMetaInfo->pTableMeta->id.uid;
 
-  assert(pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH || pExpr->tokenId == TK_IN);
+  assert(pExpr->tokenId == TK_LIKE
+         || pExpr->tokenId == TK_MATCH
+         || pExpr->tokenId == TK_NMATCH
+         || pExpr->tokenId == TK_IN);
 
-  if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH) {
+  if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH || pExpr->tokenId == TK_NMATCH) {
     char* str = taosStringBuilderGetResult(sb, NULL);
     pQueryInfo->tagCond.tbnameCond.cond = strdup(str);
     pQueryInfo->tagCond.tbnameCond.len = (int32_t) strlen(str);
@@ -8224,11 +8239,12 @@ static int32_t handleExprInHavingClause(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, S
         && pExpr->tokenId != TK_NOTNULL
         && pExpr->tokenId != TK_LIKE
         && pExpr->tokenId != TK_MATCH
+        && pExpr->tokenId != TK_NMATCH
         ) {
       return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg2);
     }
   } else {
-    if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH) {
+    if (pExpr->tokenId == TK_LIKE || pExpr->tokenId == TK_MATCH || pExpr->tokenId == TK_NMATCH) {
       return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg1);
     }
 

src/common/inc/texpr.h

@@ -34,10 +34,12 @@ struct SSchema;
 #define QUERY_COND_REL_PREFIX_IN "IN|"
 #define QUERY_COND_REL_PREFIX_LIKE "LIKE|"
 #define QUERY_COND_REL_PREFIX_MATCH "MATCH|"
+#define QUERY_COND_REL_PREFIX_NMATCH "NMATCH|"
 
 #define QUERY_COND_REL_PREFIX_IN_LEN   3
 #define QUERY_COND_REL_PREFIX_LIKE_LEN 5
 #define QUERY_COND_REL_PREFIX_MATCH_LEN 6
+#define QUERY_COND_REL_PREFIX_NMATCH_LEN 7
 
 typedef bool (*__result_filter_fn_t)(const void *, void *);
 typedef void (*__do_filter_suppl_fn_t)(void *, void *);

src/common/src/texpr.c

@@ -440,7 +440,16 @@ tExprNode* exprTreeFromTableName(const char* tbnameCond) {
     memcpy(pVal->pz, tbnameCond + QUERY_COND_REL_PREFIX_MATCH_LEN, len);
     pVal->nType = TSDB_DATA_TYPE_BINARY;
     pVal->nLen = (int32_t)len;
-
+  } else if (strncmp(tbnameCond, QUERY_COND_REL_PREFIX_NMATCH, QUERY_COND_REL_PREFIX_NMATCH_LEN) == 0) {
+    right->nodeType = TSQL_NODE_VALUE;
+    expr->_node.optr = TSDB_RELATION_NMATCH;
+    tVariant* pVal = exception_calloc(1, sizeof(tVariant));
+    right->pVal = pVal;
+    size_t len = strlen(tbnameCond + QUERY_COND_REL_PREFIX_NMATCH_LEN) + 1;
+    pVal->pz = exception_malloc(len);
+    memcpy(pVal->pz, tbnameCond + QUERY_COND_REL_PREFIX_NMATCH_LEN, len);
+    pVal->nType = TSDB_DATA_TYPE_BINARY;
+    pVal->nLen = (int32_t)len;
   } else if (strncmp(tbnameCond, QUERY_COND_REL_PREFIX_IN, QUERY_COND_REL_PREFIX_IN_LEN) == 0) {
     right->nodeType = TSQL_NODE_VALUE;
     expr->_node.optr = TSDB_RELATION_IN;

src/common/src/tglobal.c

@@ -1039,6 +1039,16 @@ static void doInitGlobalConfig(void) {
   cfg.unitType = TAOS_CFG_UTYPE_BYTE;
   taosInitConfigOption(cfg);
 
+  cfg.option = "maxRegexStringLen";
+  cfg.ptr = &tsMaxRegexStringLen;
+  cfg.valType = TAOS_CFG_VTYPE_INT32;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_CLIENT | TSDB_CFG_CTYPE_B_SHOW;
+  cfg.minValue = 0;
+  cfg.maxValue = TSDB_MAX_FIELD_LEN;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_BYTE;
+  taosInitConfigOption(cfg);
+
   cfg.option = "maxNumOfOrderedRes";
   cfg.ptr = &tsMaxNumOfOrderedResults;
   cfg.valType = TAOS_CFG_VTYPE_INT32;

src/inc/taosdef.h

@@ -165,6 +165,7 @@ do { \
 #define TSDB_RELATION_NOT         13
 
 #define TSDB_RELATION_MATCH       14
+#define TSDB_RELATION_NMATCH      15
 
 #define TSDB_BINARY_OP_ADD        30
 #define TSDB_BINARY_OP_SUBTRACT   31

src/inc/ttokendef.h

@@ -38,179 +38,180 @@
 #define TK_IS                              20
 #define TK_LIKE                            21
 #define TK_MATCH                           22
-#define TK_GLOB                            23
-#define TK_BETWEEN                         24
-#define TK_IN                              25
-#define TK_GT                              26
-#define TK_GE                              27
-#define TK_LT                              28
-#define TK_LE                              29
-#define TK_BITAND                          30
-#define TK_BITOR                           31
-#define TK_LSHIFT                          32
-#define TK_RSHIFT                          33
-#define TK_PLUS                            34
-#define TK_MINUS                           35
-#define TK_DIVIDE                          36
-#define TK_TIMES                           37
-#define TK_STAR                            38
-#define TK_SLASH                           39
-#define TK_REM                             40
-#define TK_CONCAT                          41
-#define TK_UMINUS                          42
-#define TK_UPLUS                           43
-#define TK_BITNOT                          44
-#define TK_SHOW                            45
-#define TK_DATABASES                       46
-#define TK_TOPICS                          47
-#define TK_FUNCTIONS                       48
-#define TK_MNODES                          49
-#define TK_DNODES                          50
-#define TK_ACCOUNTS                        51
-#define TK_USERS                           52
-#define TK_MODULES                         53
-#define TK_QUERIES                         54
-#define TK_CONNECTIONS                     55
-#define TK_STREAMS                         56
-#define TK_VARIABLES                       57
-#define TK_SCORES                          58
-#define TK_GRANTS                          59
-#define TK_VNODES                          60
-#define TK_DOT                             61
-#define TK_CREATE                          62
-#define TK_TABLE                           63
-#define TK_STABLE                          64
-#define TK_DATABASE                        65
-#define TK_TABLES                          66
-#define TK_STABLES                         67
-#define TK_VGROUPS                         68
-#define TK_DROP                            69
-#define TK_TOPIC                           70
-#define TK_FUNCTION                        71
-#define TK_DNODE                           72
-#define TK_USER                            73
-#define TK_ACCOUNT                         74
-#define TK_USE                             75
-#define TK_DESCRIBE                        76
-#define TK_DESC                            77
-#define TK_ALTER                           78
-#define TK_PASS                            79
-#define TK_PRIVILEGE                       80
-#define TK_LOCAL                           81
-#define TK_COMPACT                         82
-#define TK_LP                              83
-#define TK_RP                              84
-#define TK_IF                              85
-#define TK_EXISTS                          86
-#define TK_AS                              87
-#define TK_OUTPUTTYPE                      88
-#define TK_AGGREGATE                       89
-#define TK_BUFSIZE                         90
-#define TK_PPS                             91
-#define TK_TSERIES                         92
-#define TK_DBS                             93
-#define TK_STORAGE                         94
-#define TK_QTIME                           95
-#define TK_CONNS                           96
-#define TK_STATE                           97
-#define TK_COMMA                           98
-#define TK_KEEP                            99
-#define TK_CACHE                          100
-#define TK_REPLICA                        101
-#define TK_QUORUM                         102
-#define TK_DAYS                           103
-#define TK_MINROWS                        104
-#define TK_MAXROWS                        105
-#define TK_BLOCKS                         106
-#define TK_CTIME                          107
-#define TK_WAL                            108
-#define TK_FSYNC                          109
-#define TK_COMP                           110
-#define TK_PRECISION                      111
-#define TK_UPDATE                         112
-#define TK_CACHELAST                      113
-#define TK_PARTITIONS                     114
-#define TK_UNSIGNED                       115
-#define TK_TAGS                           116
-#define TK_USING                          117
-#define TK_NULL                           118
-#define TK_NOW                            119
-#define TK_SELECT                         120
-#define TK_UNION                          121
-#define TK_ALL                            122
-#define TK_DISTINCT                       123
-#define TK_FROM                           124
-#define TK_VARIABLE                       125
-#define TK_INTERVAL                       126
-#define TK_EVERY                          127
-#define TK_SESSION                        128
-#define TK_STATE_WINDOW                   129
-#define TK_FILL                           130
-#define TK_SLIDING                        131
-#define TK_ORDER                          132
-#define TK_BY                             133
-#define TK_ASC                            134
-#define TK_GROUP                          135
-#define TK_HAVING                         136
-#define TK_LIMIT                          137
-#define TK_OFFSET                         138
-#define TK_SLIMIT                         139
-#define TK_SOFFSET                        140
-#define TK_WHERE                          141
-#define TK_RESET                          142
-#define TK_QUERY                          143
-#define TK_SYNCDB                         144
-#define TK_ADD                            145
-#define TK_COLUMN                         146
-#define TK_MODIFY                         147
-#define TK_TAG                            148
-#define TK_CHANGE                         149
-#define TK_SET                            150
-#define TK_KILL                           151
-#define TK_CONNECTION                     152
-#define TK_STREAM                         153
-#define TK_COLON                          154
-#define TK_ABORT                          155
-#define TK_AFTER                          156
-#define TK_ATTACH                         157
-#define TK_BEFORE                         158
-#define TK_BEGIN                          159
-#define TK_CASCADE                        160
-#define TK_CLUSTER                        161
-#define TK_CONFLICT                       162
-#define TK_COPY                           163
-#define TK_DEFERRED                       164
-#define TK_DELIMITERS                     165
-#define TK_DETACH                         166
-#define TK_EACH                           167
-#define TK_END                            168
-#define TK_EXPLAIN                        169
-#define TK_FAIL                           170
-#define TK_FOR                            171
-#define TK_IGNORE                         172
-#define TK_IMMEDIATE                      173
-#define TK_INITIALLY                      174
-#define TK_INSTEAD                        175
-#define TK_KEY                            176
-#define TK_OF                             177
-#define TK_RAISE                          178
-#define TK_REPLACE                        179
-#define TK_RESTRICT                       180
-#define TK_ROW                            181
-#define TK_STATEMENT                      182
-#define TK_TRIGGER                        183
-#define TK_VIEW                           184
-#define TK_IPTOKEN                        185
-#define TK_SEMI                           186
-#define TK_NONE                           187
-#define TK_PREV                           188
-#define TK_LINEAR                         189
-#define TK_IMPORT                         190
-#define TK_TBNAME                         191
-#define TK_JOIN                           192
-#define TK_INSERT                         193
-#define TK_INTO                           194
-#define TK_VALUES                         195
+#define TK_NMATCH                          23
+#define TK_GLOB                            24
+#define TK_BETWEEN                         25
+#define TK_IN                              26
+#define TK_GT                              27
+#define TK_GE                              28
+#define TK_LT                              29
+#define TK_LE                              30
+#define TK_BITAND                          31
+#define TK_BITOR                           32
+#define TK_LSHIFT                          33
+#define TK_RSHIFT                          34
+#define TK_PLUS                            35
+#define TK_MINUS                           36
+#define TK_DIVIDE                          37
+#define TK_TIMES                           38
+#define TK_STAR                            39
+#define TK_SLASH                           40
+#define TK_REM                             41
+#define TK_CONCAT                          42
+#define TK_UMINUS                          43
+#define TK_UPLUS                           44
+#define TK_BITNOT                          45
+#define TK_SHOW                            46
+#define TK_DATABASES                       47
+#define TK_TOPICS                          48
+#define TK_FUNCTIONS                       49
+#define TK_MNODES                          50
+#define TK_DNODES                          51
+#define TK_ACCOUNTS                        52
+#define TK_USERS                           53
+#define TK_MODULES                         54
+#define TK_QUERIES                         55
+#define TK_CONNECTIONS                     56
+#define TK_STREAMS                         57
+#define TK_VARIABLES                       58
+#define TK_SCORES                          59
+#define TK_GRANTS                          60
+#define TK_VNODES                          61
+#define TK_DOT                             62
+#define TK_CREATE                          63
+#define TK_TABLE                           64
+#define TK_STABLE                          65
+#define TK_DATABASE                        66
+#define TK_TABLES                          67
+#define TK_STABLES                         68
+#define TK_VGROUPS                         69
+#define TK_DROP                            70
+#define TK_TOPIC                           71
+#define TK_FUNCTION                        72
+#define TK_DNODE                           73
+#define TK_USER                            74
+#define TK_ACCOUNT                         75
+#define TK_USE                             76
+#define TK_DESCRIBE                        77
+#define TK_DESC                            78
+#define TK_ALTER                           79
+#define TK_PASS                            80
+#define TK_PRIVILEGE                       81
+#define TK_LOCAL                           82
+#define TK_COMPACT                         83
+#define TK_LP                              84
+#define TK_RP                              85
+#define TK_IF                              86
+#define TK_EXISTS                          87
+#define TK_AS                              88
+#define TK_OUTPUTTYPE                      89
+#define TK_AGGREGATE                       90
+#define TK_BUFSIZE                         91
+#define TK_PPS                             92
+#define TK_TSERIES                         93
+#define TK_DBS                             94
+#define TK_STORAGE                         95
+#define TK_QTIME                           96
+#define TK_CONNS                           97
+#define TK_STATE                           98
+#define TK_COMMA                           99
+#define TK_KEEP                           100
+#define TK_CACHE                          101
+#define TK_REPLICA                        102
+#define TK_QUORUM                         103
+#define TK_DAYS                           104
+#define TK_MINROWS                        105
+#define TK_MAXROWS                        106
+#define TK_BLOCKS                         107
+#define TK_CTIME                          108
+#define TK_WAL                            109
+#define TK_FSYNC                          110
+#define TK_COMP                           111
+#define TK_PRECISION                      112
+#define TK_UPDATE                         113
+#define TK_CACHELAST                      114
+#define TK_PARTITIONS                     115
+#define TK_UNSIGNED                       116
+#define TK_TAGS                           117
+#define TK_USING                          118
+#define TK_NULL                           119
+#define TK_NOW                            120
+#define TK_SELECT                         121
+#define TK_UNION                          122
+#define TK_ALL                            123
+#define TK_DISTINCT                       124
+#define TK_FROM                           125
+#define TK_VARIABLE                       126
+#define TK_INTERVAL                       127
+#define TK_EVERY                          128
+#define TK_SESSION                        129
+#define TK_STATE_WINDOW                   130
+#define TK_FILL                           131
+#define TK_SLIDING                        132
+#define TK_ORDER                          133
+#define TK_BY                             134
+#define TK_ASC                            135
+#define TK_GROUP                          136
+#define TK_HAVING                         137
+#define TK_LIMIT                          138
+#define TK_OFFSET                         139
+#define TK_SLIMIT                         140
+#define TK_SOFFSET                        141
+#define TK_WHERE                          142
+#define TK_RESET                          143
+#define TK_QUERY                          144
+#define TK_SYNCDB                         145
+#define TK_ADD                            146
+#define TK_COLUMN                         147
+#define TK_MODIFY                         148
+#define TK_TAG                            149
+#define TK_CHANGE                         150
+#define TK_SET                            151
+#define TK_KILL                           152
+#define TK_CONNECTION                     153
+#define TK_STREAM                         154
+#define TK_COLON                          155
+#define TK_ABORT                          156
+#define TK_AFTER                          157
+#define TK_ATTACH                         158
+#define TK_BEFORE                         159
+#define TK_BEGIN                          160
+#define TK_CASCADE                        161
+#define TK_CLUSTER                        162
+#define TK_CONFLICT                       163
+#define TK_COPY                           164
+#define TK_DEFERRED                       165
+#define TK_DELIMITERS                     166
+#define TK_DETACH                         167
+#define TK_EACH                           168
+#define TK_END                            169
+#define TK_EXPLAIN                        170
+#define TK_FAIL                           171
+#define TK_FOR                            172
+#define TK_IGNORE                         173
+#define TK_IMMEDIATE                      174
+#define TK_INITIALLY                      175
+#define TK_INSTEAD                        176
+#define TK_KEY                            177
+#define TK_OF                             178
+#define TK_RAISE                          179
+#define TK_REPLACE                        180
+#define TK_RESTRICT                       181
+#define TK_ROW                            182
+#define TK_STATEMENT                      183
+#define TK_TRIGGER                        184
+#define TK_VIEW                           185
+#define TK_IPTOKEN                        186
+#define TK_SEMI                           187
+#define TK_NONE                           188
+#define TK_PREV                           189
+#define TK_LINEAR                         190
+#define TK_IMPORT                         191
+#define TK_TBNAME                         192
+#define TK_JOIN                           193
+#define TK_INSERT                         194
+#define TK_INTO                           195
+#define TK_VALUES                         196
 
 
 #define TK_SPACE                          300

src/query/inc/sql.y

@@ -11,7 +11,7 @@
 %left OR.
 %left AND.
 %right NOT.
-%left EQ NE ISNULL NOTNULL IS LIKE MATCH GLOB BETWEEN IN.
+%left EQ NE ISNULL NOTNULL IS LIKE MATCH NMATCH GLOB BETWEEN IN.
 %left GT GE LT LE.
 %left BITAND BITOR LSHIFT RSHIFT.
 %left PLUS MINUS.
@@ -753,6 +753,7 @@ expr(A) ::= expr(X) LIKE expr(Y).    {A = tSqlExprCreate(X, Y, TK_LIKE);  }
 
 // match expression
 expr(A) ::= expr(X) MATCH expr(Y).    {A = tSqlExprCreate(X, Y, TK_MATCH);  }
+expr(A) ::= expr(X) NMATCH expr(Y).    {A = tSqlExprCreate(X, Y, TK_NMATCH);  }
 
 //in expression
 expr(A) ::= expr(X) IN LP exprlist(Y) RP.   {A = tSqlExprCreate(X, (tSqlExpr*)Y, TK_IN); }
@@ -919,5 +920,5 @@ cmd ::= KILL QUERY INTEGER(X) COLON(Z) INTEGER(Y).        {X.n += (Z.n + Y.n); s
 
 %fallback ID ABORT AFTER ASC ATTACH BEFORE BEGIN CASCADE CLUSTER CONFLICT COPY DATABASE DEFERRED
   DELIMITERS DESC DETACH EACH END EXPLAIN FAIL FOR GLOB IGNORE IMMEDIATE INITIALLY INSTEAD
-  LIKE MATCH KEY OF OFFSET RAISE REPLACE RESTRICT ROW STATEMENT TRIGGER VIEW ALL
+  LIKE MATCH NMATCH KEY OF OFFSET RAISE REPLACE RESTRICT ROW STATEMENT TRIGGER VIEW ALL
   NOW IPTOKEN SEMI NONE PREV LINEAR IMPORT TBNAME JOIN STABLE NULL INSERT INTO VALUES.

src/query/src/qFilter.c

@@ -29,6 +29,7 @@ OptrStr gOptrStr[] = {
   {TSDB_RELATION_NOT_EQUAL,                "!="},
   {TSDB_RELATION_LIKE,                     "like"},
   {TSDB_RELATION_MATCH,                    "match"},
+  {TSDB_RELATION_MATCH,                    "nmatch"},
   {TSDB_RELATION_ISNULL,                   "is null"},
   {TSDB_RELATION_NOTNULL,                  "not null"},
   {TSDB_RELATION_IN,                       "in"},
@@ -157,7 +158,7 @@ int8_t filterGetRangeCompFuncFromOptrs(uint8_t optr, uint8_t optr2) {
 __compar_fn_t gDataCompare[] = {compareInt32Val, compareInt8Val, compareInt16Val, compareInt64Val, compareFloatVal,
   compareDoubleVal, compareLenPrefixedStr, compareStrPatternComp, compareFindItemInSet, compareWStrPatternComp, 
   compareLenPrefixedWStr, compareUint8Val, compareUint16Val, compareUint32Val, compareUint64Val,
-  setCompareBytes1, setCompareBytes2, setCompareBytes4, setCompareBytes8, compareStrRegexComp,
+  setCompareBytes1, setCompareBytes2, setCompareBytes4, setCompareBytes8, compareStrRegexCompMatch, compareStrRegexCompNMatch
 };
 
 int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) {
@@ -198,6 +199,8 @@ int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) {
     case TSDB_DATA_TYPE_BINARY: {
       if (optr == TSDB_RELATION_MATCH) {
         comparFn = 19;
+      } else if (optr == TSDB_RELATION_NMATCH) {
+        comparFn = 20;
       } else if (optr == TSDB_RELATION_LIKE) { /* wildcard query using like operator */
         comparFn = 7;
       } else if (optr == TSDB_RELATION_IN) {
@@ -212,6 +215,8 @@ int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) {
     case TSDB_DATA_TYPE_NCHAR: {
       if (optr == TSDB_RELATION_MATCH) {
         comparFn = 19;
+      } else if (optr == TSDB_RELATION_NMATCH) {
+        comparFn = 20;
       } else if (optr == TSDB_RELATION_LIKE) {
         comparFn = 9;
       } else if (optr == TSDB_RELATION_IN) {
@@ -1879,6 +1884,9 @@ bool filterDoCompare(__compar_fn_t func, uint8_t optr, void *left, void *right)
     case TSDB_RELATION_MATCH: {
       return ret == 0;
     }
+    case TSDB_RELATION_NMATCH: {
+      return ret == 0;
+    }
     case TSDB_RELATION_IN: {
       return ret == 1;
     }

src/tsdb/src/tsdbRead.c

@@ -3708,6 +3708,9 @@ static bool tableFilterFp(const void* pNode, void* param) {
     case TSDB_RELATION_MATCH: {
       return ret == 0;
     }
+    case TSDB_RELATION_NMATCH: {
+      return ret == 0;
+    }
     case TSDB_RELATION_IN: {
       return ret == 1;
     }
@@ -4043,6 +4046,8 @@ static int32_t setQueryCond(tQueryInfo *queryColInfo, SQueryCond* pCond) {
     assert(0);
   } else if (optr == TSDB_RELATION_MATCH) {
     assert(0);
+  } else if (optr == TSDB_RELATION_NMATCH) {
+    assert(0);
   }
 
   return TSDB_CODE_SUCCESS;
@@ -4200,7 +4205,9 @@ static void queryIndexlessColumn(SSkipList* pSkipList, tQueryInfo* pQueryInfo, S
     if (pQueryInfo->sch.colId == TSDB_TBNAME_COLUMN_INDEX) {
       if (pQueryInfo->optr == TSDB_RELATION_IN) {
         addToResult = pQueryInfo->compare(name, pQueryInfo->q);
-      } else if (pQueryInfo->optr == TSDB_RELATION_LIKE || pQueryInfo->optr == TSDB_RELATION_MATCH) {
+      } else if (pQueryInfo->optr == TSDB_RELATION_LIKE ||
+                 pQueryInfo->optr == TSDB_RELATION_MATCH ||
+                 pQueryInfo->optr == TSDB_RELATION_NMATCH) {
         addToResult = !pQueryInfo->compare(name, pQueryInfo->q);
       }
     } else {
@@ -4232,7 +4239,8 @@ void getTableListfromSkipList(tExprNode *pExpr, SSkipList *pSkipList, SArray *re
     param->setupInfoFn(pExpr, param->pExtInfo);
 
     tQueryInfo *pQueryInfo = pExpr->_node.info;
-    if (pQueryInfo->indexed && (pQueryInfo->optr != TSDB_RELATION_LIKE && pQueryInfo->optr != TSDB_RELATION_MATCH
+    if (pQueryInfo->indexed && (pQueryInfo->optr != TSDB_RELATION_LIKE
+                                && pQueryInfo->optr != TSDB_RELATION_MATCH && pQueryInfo->optr != TSDB_RELATION_NMATCH
                                 && pQueryInfo->optr != TSDB_RELATION_IN)) {
       queryIndexedColumn(pSkipList, pQueryInfo, result);
     } else {

src/util/inc/tcompare.h

@@ -84,6 +84,8 @@ int32_t compareLenPrefixedStr(const void *pLeft, const void *pRight);
 int32_t compareLenPrefixedWStr(const void *pLeft, const void *pRight);
 int32_t compareStrPatternComp(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 compareFindItemInSet(const void *pLeft, const void* pRight);
 int32_t compareWStrPatternComp(const void* pLeft, const void* pRight);
 

src/util/inc/tconfig.h

@@ -20,7 +20,7 @@
 extern "C" {
 #endif
 
-#define TSDB_CFG_MAX_NUM    122
+#define TSDB_CFG_MAX_NUM    123
 #define TSDB_CFG_PRINT_LEN  23
 #define TSDB_CFG_OPTION_LEN 24
 #define TSDB_CFG_VALUE_LEN  41

src/util/src/tcompare.c

@@ -350,6 +350,14 @@ int32_t compareStrPatternComp(const void* pLeft, const void* pRight) {
   return (ret == TSDB_PATTERN_MATCH) ? 0 : 1;
 }
 
+int32_t compareStrRegexCompMatch(const void* pLeft, const void* pRight) {
+  return compareStrRegexComp(pLeft, pRight);
+}
+
+int32_t compareStrRegexCompNMatch(const void* pLeft, const void* pRight) {
+  return compareStrRegexComp(pLeft, pRight) ? 0 : 1;
+}
+
 int32_t compareStrRegexComp(const void* pLeft, const void* pRight) {
   size_t sz = varDataLen(pRight);
   char *pattern = malloc(sz + 1);
@@ -449,7 +457,9 @@ __compar_fn_t getComparFunc(int32_t type, int32_t optr) {
     case TSDB_DATA_TYPE_DOUBLE:    comparFn = compareDoubleVal; break;
     case TSDB_DATA_TYPE_BINARY: {
       if (optr == TSDB_RELATION_MATCH) {
-        comparFn = compareStrRegexComp;
+        comparFn = compareStrRegexCompMatch;
+      } else if (optr == TSDB_RELATION_NMATCH) {
+        comparFn = compareStrRegexCompNMatch;
       } else if (optr == TSDB_RELATION_LIKE) { /* wildcard query using like operator */
         comparFn = compareStrPatternComp;
       } else if (optr == TSDB_RELATION_IN) {
@@ -463,7 +473,9 @@ __compar_fn_t getComparFunc(int32_t type, int32_t optr) {
 
     case TSDB_DATA_TYPE_NCHAR: {
       if (optr == TSDB_RELATION_MATCH) {
-        comparFn = compareStrRegexComp;
+        comparFn = compareStrRegexCompMatch;
+      } else if (optr == TSDB_RELATION_NMATCH) {
+        comparFn = compareStrRegexCompNMatch;
       } else if (optr == TSDB_RELATION_LIKE) {
         comparFn = compareWStrPatternComp;
       } else if (optr == TSDB_RELATION_IN) {

src/util/src/ttokenizer.c

@@ -195,6 +195,7 @@ static SKeyword keywordTable[] = {
     {"INITIALLY",    TK_INITIALLY},
     {"INSTEAD",      TK_INSTEAD},
     {"MATCH",        TK_MATCH},
+    {"NMATCH",       TK_NMATCH},
     {"KEY",          TK_KEY},
     {"OF",           TK_OF},
     {"RAISE",        TK_RAISE},

tests/script/general/parser/regex.sim

@@ -29,13 +29,22 @@ endi
 
 
 sql select tbname from $st_name where tbname match '^ct[[:digit:]]'
-
 if $rows != 2 then
   return -1
 endi
 
+sql select tbname from $st_name where tbname nmatch '^ct[[:digit:]]'
+if $rows != 1 then
+  return -1
+endi
+
 sql select tbname from $st_name where tbname match '.*'
-if $rows !=3 then
+if $rows != 3 then
+  return -1
+endi
+
+sql select tbname from $st_name where tbname nmatch '.*'
+if $rows != 0 then
   return -1
 endi
 
@@ -44,6 +53,11 @@ if $rows != 2 then
   return -1
 endi
 
+sql select tbname from $st_name where t1b nmatch '[[:lower:]]+'
+if $rows != 1 then
+  return -1
+endi
+
 sql insert into $ct1_name values(now, 'this is engine')
 sql insert into $ct2_name values(now, 'this is app egnine')
 
@@ -56,6 +70,52 @@ if $rows != 1 then
   return -1
 endi  
 
+sql select c1b from $st_name where c1b nmatch 'engine'
+if $data00 != @this is app egnine@ then
+  return -1
+endi
+
+if $rows != 1 then
+  return -1
+endi  
+
+sql_error select c1b from $st_name where c1b match e;
+sql_error select c1b from $st_name where c1b nmatch e;
+
+sql create table wrong_type(ts timestamp, c0 tinyint, c1 smallint, c2 int, c3 bigint, c4 float, c5 double, c6 bool, c7 nchar(20)) tags(t0 tinyint, t1 smallint, t2 int, t3 bigint, t4 float, t5 double, t6 bool, t7 nchar(10))
+sql insert into wrong_type_1 using wrong_type tags(1, 2, 3, 4, 5, 6, true, 'notsupport') values(now, 1, 2, 3, 4, 5, 6, false, 'notsupport') 
+sql_error select * from wrong_type where ts match '.*'
+sql_error select * from wrong_type where ts nmatch '.*'
+sql_error select * from wrong_type where c0 match '.*'
+sql_error select * from wrong_type where c0 nmatch '.*'
+sql_error select * from wrong_type where c1 match '.*'
+sql_error select * from wrong_type where c1 nmatch '.*'
+sql_error select * from wrong_type where c2 match '.*'
+sql_error select * from wrong_type where c2 nmatch '.*'
+sql_error select * from wrong_type where c3 match '.*'
+sql_error select * from wrong_type where c3 nmatch '.*'
+sql_error select * from wrong_type where c4 match '.*'
+sql_error select * from wrong_type where c4 nmatch '.*'
+sql_error select * from wrong_type where c5 match '.*'
+sql_error select * from wrong_type where c5 nmatch '.*'
+sql_error select * from wrong_type where c6 match '.*'
+sql_error select * from wrong_type where c6 nmatch '.*'
+sql_error select * from wrong_type where c7 match '.*'
+sql_error select * from wrong_type where c7 nmatch '.*'
+sql_error select * from wrong_type where t1 match '.*'
+sql_error select * from wrong_type where t1 nmatch '.*'
+sql_error select * from wrong_type where t2 match '.*'
+sql_error select * from wrong_type where t2 nmatch '.*'
+sql_error select * from wrong_type where t3 match '.*'
+sql_error select * from wrong_type where t3 nmatch '.*'
+sql_error select * from wrong_type where t4 match '.*'
+sql_error select * from wrong_type where t4 nmatch '.*'
+sql_error select * from wrong_type where t5 match '.*'
+sql_error select * from wrong_type where t5 nmatch '.*'
+sql_error select * from wrong_type where t6 match '.*'
+sql_error select * from wrong_type where t6 nmatch '.*'
+sql_error select * from wrong_type where t7 match '.*'
+sql_error select * from wrong_type where t7 nmatch '.*'
 
 system sh/exec.sh -n dnode1 -s stop -x SIGINT