fix:[TS-5761] error for in operator

source/libs/scalar/src/filter.c

@@ -4848,7 +4848,6 @@ EDealRes fltReviseRewriter(SNode **pNode, void *pContext) {
       }
 
       SColumnNode *refNode = (SColumnNode *)node->pLeft;
-      int32_t      type = refNode->node.resType.type;
       SExprNode   *exprNode = NULL;
       if (OP_TYPE_IN != node->opType && OP_TYPE_NOT_IN != node->opType) {
         SValueNode  *valueNode = (SValueNode *)node->pRight;
@@ -4857,29 +4856,34 @@ EDealRes fltReviseRewriter(SNode **pNode, void *pContext) {
           valueNode->node.resType.type = TSDB_DATA_TYPE_BIGINT;
         }
         exprNode = &valueNode->node;
-        type = vectorGetConvertType(refNode->node.resType.type, exprNode->resType.type);
+        int32_t type = vectorGetConvertType(refNode->node.resType.type, exprNode->resType.type);
+        if (0 != type && type != refNode->node.resType.type) {
+          stat->scalarMode = true;
+        }
+        return DEAL_RES_CONTINUE;
       } else {
         SNodeListNode *listNode = (SNodeListNode *)node->pRight;
-        if (LIST_LENGTH(listNode->pNodeList) > 10) {
+        if (LIST_LENGTH(listNode->pNodeList) > 10 || OP_TYPE_NOT_IN == node->opType) {
           stat->scalarMode = true;
-          return DEAL_RES_CONTINUE;
         }
+        int32_t type = -1;
         exprNode = &listNode->node;
         SListCell     *cell = listNode->pNodeList->pHead;
         for (int32_t i = 0; i < listNode->pNodeList->length; ++i) {
           SValueNode *valueNode = (SValueNode *)cell->pNode;
           cell = cell->pNext;
-          int32_t tmp = vectorGetConvertType(type, valueNode->node.resType.type);
+          int32_t tmp = vectorGetConvertType(refNode->node.resType.type, valueNode->node.resType.type);
           if (tmp != 0){
+            stat->scalarMode = true;
+            if (IS_NUMERIC_TYPE(tmp) && tmp > type){
               type = tmp;
             }
           }
+
+        }
         if (IS_NUMERIC_TYPE(type)){
           exprNode->resType.type = type;
         }
-      }
-      if ((0 != type && type != refNode->node.resType.type) || OP_TYPE_NOT_IN == node->opType) {
-        stat->scalarMode = true;
         return DEAL_RES_CONTINUE;
       }
     }

source/libs/scalar/src/scalar.c

@@ -201,9 +201,6 @@ int32_t scalarGenerateSetFromList(void **data, void *pNode, uint32_t type, int8_
     cell = cell->pNext;
   }
 
-  if (taosHashGetSize(pObj) == 0) {
-    goto _return;
-  }
   *data = pObj;
 
   colDataDestroy(out.columnData);
@@ -386,8 +383,21 @@ int32_t sclInitParam(SNode *node, SScalarParam *param, SScalarCtx *ctx, int32_t
         SCL_RET(TSDB_CODE_QRY_INVALID_INPUT);
       }
 
-      int32_t type = ctx->type.peerType;
+      int32_t type = -1;
-      if (IS_VAR_DATA_TYPE(ctx->type.selfType) && IS_NUMERIC_TYPE(ctx->type.peerType)){
+      SListCell     *cell = nodeList->pNodeList->pHead;
+      for (int32_t i = 0; i < nodeList->pNodeList->length; ++i) {
+        SValueNode *valueNode = (SValueNode *)cell->pNode;
+        cell = cell->pNext;
+        int32_t tmp = vectorGetConvertType(ctx->type.selfType, valueNode->node.resType.type);
+        if (tmp != 0 && IS_NUMERIC_TYPE(tmp) && tmp > type){
+            type = tmp;
+        }
+      }
+      if (IS_NUMERIC_TYPE(type)){
+        ctx->type.peerType = type;
+      }
+      type = ctx->type.peerType;
+      if (IS_VAR_DATA_TYPE(ctx->type.selfType) && IS_NUMERIC_TYPE(type)){
         SCL_ERR_RET(scalarGenerateSetFromList((void **)&param->pHashFilter, node, type, 1));
         SCL_ERR_RET(scalarGenerateSetFromList((void **)&param->pHashFilterVar, node, ctx->type.selfType, 2));
       } else {

source/libs/scalar/src/sclvector.c

@@ -2013,9 +2013,17 @@ int32_t doVectorCompare(SScalarParam *pLeft, SScalarParam *pLeftVar, SScalarPara
 
       char *pLeftData = colDataGetData(pLeft->columnData, i);
       bool  res = filterDoCompare(fp, optr, pLeftData, pRight->pHashFilter);
-      if (pLeftVar != NULL && !res){
+      if (pLeftVar != NULL && taosHashGetSize(pRight->pHashFilterVar) > 0){
+        do{
+          if (optr == OP_TYPE_IN && res){
+            break;
+          }
+          if (optr == OP_TYPE_NOT_IN && !res){
+            break;
+          }
           pLeftData = colDataGetData(pLeftVar->columnData, i);
-        res = res || filterDoCompare(fpVar, optr, pLeftData, pRight->pHashFilterVar);
+          res = filterDoCompare(fpVar, optr, pLeftData, pRight->pHashFilterVar);
+        }while(0);
       }
       colDataSetInt8(pOut->columnData, i, (int8_t *)&res);
       if (res) {

source/util/src/thashutil.c

@@ -152,8 +152,8 @@ uint32_t taosFloatHash(const char *key, uint32_t UNUSED_PARAM(len)) {
   if (FLT_EQUAL(f, 0.0)) {
     return 0;
   }
-  if (fabs(f) < FLT_MAX / BASE - DLT) {
+  if (fabs(f) < INT32_MAX) {
-    int32_t t = (int32_t)(round(BASE * (f + DLT)));
+    int32_t t = (int32_t)(floor(f));
     return (uint32_t)t;
   } else {
     return 0x7fc00000;
@@ -168,8 +168,8 @@ uint32_t taosDoubleHash(const char *key, uint32_t UNUSED_PARAM(len)) {
   if (FLT_EQUAL(f, 0.0)) {
     return 0;
   }
-  if (fabs(f) < DBL_MAX / BASE - DLT) {
+  if (fabs(f) < INT32_MAX) {
-    int32_t t = (int32_t)(round(BASE * (f + DLT)));
+    int32_t t = (int32_t)(floor(f));
     return (uint32_t)t;
   } else {
     return 0x7fc00000;

tests/script/tsim/scalar/in.sim

@@ -44,7 +44,7 @@ if $data20 != @      Time Range: [-9223372036854775808, 9223372036854775807]@ th
 endi
 
 sql select * from tb1 where fbool in (0, 3);
-if $rows != 5 then
+if $rows != 3 then
   return -1
 endi
 
@@ -69,7 +69,7 @@ if $rows != 10 then
 endi
 
 sql select * from st1 where tbool in (0, 3);
-if $rows != 15 then
+if $rows != 5 then
   return -1
 endi
 

tests/system-test/2-query/ts-5761-scalemode.py

@@ -21,12 +21,17 @@ class TDTestCase:
         tdSql.execute(f"use db")
 
         # super tableUNSIGNED
-        tdSql.execute("CREATE TABLE t1( time TIMESTAMP, c1 BIGINT, c2 smallint, c3 double, c4 int UNSIGNED, c5 bool, c6 binary(32), c7 nchar(32));")
+        tdSql.execute("CREATE TABLE st( time TIMESTAMP, c1 BIGINT, c2 smallint, c3 double, c4 int UNSIGNED, c5 bool, c6 binary(32), c7 nchar(32)) tags(t1 binary(32), t2 nchar(32))")
+        tdSql.execute("create table t1 using st tags('1', '1.7')")
+        tdSql.execute("create table t2 using st tags('0', '')")
+        tdSql.execute("create table t3 using st tags('1', 'er')")
 
         # create index for all tags
         tdSql.execute("INSERT INTO t1 VALUES (1641024000000, 1, 1, 1, 1, 1, '1', '1.7')")
         tdSql.execute("INSERT INTO t1 VALUES (1641024000001, 0, 0, 1.7, 0, 0, '0', '')")
         tdSql.execute("INSERT INTO t1 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
+        tdSql.execute("INSERT INTO t2 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
+        tdSql.execute("INSERT INTO t3 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
 
     def check(self):
         tdSql.query(f"SELECT * FROM t1 WHERE c1 = 1.7")
@@ -84,6 +89,8 @@ class TDTestCase:
         tdSql.checkRows(1)
         tdSql.query(f"SELECT * FROM t1 WHERE c6 not in (0, 2)")
         tdSql.checkRows(2)
+        tdSql.query(f"SELECT * FROM t1 WHERE c6 not in (0, 2, 'sef')")
+        tdSql.checkRows(2)
 
         tdSql.query(f"SELECT * FROM t1 WHERE c7 = 1.7")
         tdSql.checkRows(1)
@@ -106,6 +113,18 @@ class TDTestCase:
         tdSql.query(f"SELECT * FROM t1 WHERE c7 not in ('', 2)")
         tdSql.checkRows(2)
 
+        tdSql.query(f"SELECT * FROM st WHERE t2 in ('', 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t2 not in ('', 2)")
+        tdSql.checkRows(4)
+
+        tdSql.query(f"SELECT * FROM st WHERE t1 in ('d343', 0, 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t1 in (0, 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t1 not in (0, 2)")
+        tdSql.checkRows(4)
+
     def run(self):
         self.prepareData()
         self.check()

tests/system-test/2-query/ts-5761.py

@@ -20,12 +20,17 @@ class TDTestCase:
         tdSql.execute(f"use db")
 
         # super tableUNSIGNED
-        tdSql.execute("CREATE TABLE t1( time TIMESTAMP, c1 BIGINT, c2 smallint, c3 double, c4 int UNSIGNED, c5 bool, c6 binary(32), c7 nchar(32));")
+        tdSql.execute("CREATE TABLE st( time TIMESTAMP, c1 BIGINT, c2 smallint, c3 double, c4 int UNSIGNED, c5 bool, c6 binary(32), c7 nchar(32)) tags(t1 binary(32), t2 nchar(32))")
+        tdSql.execute("create table t1 using st tags('1', '1.7')")
+        tdSql.execute("create table t2 using st tags('0', '')")
+        tdSql.execute("create table t3 using st tags('1', 'er')")
 
         # create index for all tags
         tdSql.execute("INSERT INTO t1 VALUES (1641024000000, 1, 1, 1, 1, 1, '1', '1.7')")
         tdSql.execute("INSERT INTO t1 VALUES (1641024000001, 0, 0, 1.7, 0, 0, '0', '')")
         tdSql.execute("INSERT INTO t1 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
+        tdSql.execute("INSERT INTO t2 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
+        tdSql.execute("INSERT INTO t3 VALUES (1641024000002, 1, 1, 1, 1, 1, '1', 'er')")
 
     def check(self):
         tdSql.query(f"SELECT * FROM t1 WHERE c1 = 1.7")
@@ -83,6 +88,8 @@ class TDTestCase:
         tdSql.checkRows(1)
         tdSql.query(f"SELECT * FROM t1 WHERE c6 not in (0, 2)")
         tdSql.checkRows(2)
+        tdSql.query(f"SELECT * FROM t1 WHERE c6 not in (0, 2, 'sef')")
+        tdSql.checkRows(2)
 
         tdSql.query(f"SELECT * FROM t1 WHERE c7 = 1.7")
         tdSql.checkRows(1)
@@ -105,6 +112,18 @@ class TDTestCase:
         tdSql.query(f"SELECT * FROM t1 WHERE c7 not in ('', 2)")
         tdSql.checkRows(2)
 
+        tdSql.query(f"SELECT * FROM st WHERE t2 in ('', 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t2 not in ('', 2)")
+        tdSql.checkRows(4)
+
+        tdSql.query(f"SELECT * FROM st WHERE t1 in ('d343', 0, 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t1 in (0, 2)")
+        tdSql.checkRows(1)
+        tdSql.query(f"SELECT * FROM st WHERE t1 not in (0, 2)")
+        tdSql.checkRows(4)
+
     def run(self):
         self.prepareData()
         self.check()