enh: enhance in operator

include/libs/scalar/scalar.h

@@ -42,7 +42,7 @@ int32_t scalarGetOperatorParamNum(EOperatorType type);
 int32_t scalarGenerateSetFromList(void **data, void *pNode, uint32_t type);
 
 int32_t vectorGetConvertType(int32_t type1, int32_t type2);
-int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut);
+int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut, int32_t* overflow);
 
 /* Math functions */
 int32_t absFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);

source/libs/scalar/inc/sclInt.h

@@ -56,7 +56,7 @@ typedef struct SScalarCtx {
 #define SCL_RET(c) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { terrno = _code; } return _code; } while (0)
 #define SCL_ERR_JRET(c) do { code = c; if (code != TSDB_CODE_SUCCESS) { terrno = code; goto _return; } } while (0)
 
-int32_t doConvertDataType(SValueNode* pValueNode, SScalarParam* out);
+int32_t doConvertDataType(SValueNode* pValueNode, SScalarParam* out, int32_t* overflow);
 int32_t sclCreateColumnInfoData(SDataType* pType, int32_t numOfRows, SScalarParam* pParam);
 int32_t sclConvertToTsValueNode(int8_t precision, SValueNode* valueNode);
 

source/libs/scalar/inc/sclvector.h

@@ -94,7 +94,7 @@ static FORCE_INLINE _getDoubleValue_fn_t getVectorDoubleValueFn(int32_t srcType)
   return p;
 }
 
-typedef void (*_bufConverteFunc)(char *buf, SScalarParam* pOut, int32_t outType);
+typedef void (*_bufConverteFunc)(char *buf, SScalarParam* pOut, int32_t outType, int32_t* overflow);
 typedef void (*_bin_scalar_fn_t)(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *output, int32_t order);
 _bin_scalar_fn_t getBinScalarOperatorFn(int32_t binOperator);
 

source/libs/scalar/src/filter.c

@@ -1042,12 +1042,18 @@ int32_t fltAddGroupUnitFromNode(SFilterInfo *info, SNode* tree, SArray *group) {
     
     for (int32_t i = 0; i < listNode->pNodeList->length; ++i) {
       SValueNode *valueNode = (SValueNode *)cell->pNode;
-      if (valueNode->node.resType.type != type) {
-        code = doConvertDataType(valueNode, &out);
+      if (valueNode->node.resType.type != type) {        
+        int32_t overflow = 0;
+        code = doConvertDataType(valueNode, &out, &overflow);
         if (code) {
   //        fltError("convert from %d to %d failed", in.type, out.type);
           FLT_ERR_RET(code);
         }
+
+        if (overflow) {
+          cell = cell->pNext;
+          continue;
+        }
         
         len = tDataTypes[type].bytes;
 
@@ -1835,7 +1841,7 @@ int32_t fltInitValFieldData(SFilterInfo *info) {
       }
 
       // todo refactor the convert
-      int32_t code = doConvertDataType(var, &out);
+      int32_t code = doConvertDataType(var, &out, NULL);
       if (code != TSDB_CODE_SUCCESS) {
         qError("convert value to type[%d] failed", type);
         return TSDB_CODE_TSC_INVALID_OPERATION;

source/libs/scalar/src/scalar.c

@@ -59,7 +59,7 @@ int32_t sclCreateColumnInfoData(SDataType* pType, int32_t numOfRows, SScalarPara
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t doConvertDataType(SValueNode* pValueNode, SScalarParam* out) {
+int32_t doConvertDataType(SValueNode* pValueNode, SScalarParam* out, int32_t* overflow) {
   SScalarParam in = {.numOfRows = 1};
   int32_t code = sclCreateColumnInfoData(&pValueNode->node.resType, 1, &in);
   if (code != TSDB_CODE_SUCCESS) {
@@ -69,7 +69,7 @@ int32_t doConvertDataType(SValueNode* pValueNode, SScalarParam* out) {
   colDataAppend(in.columnData, 0, nodesGetValueFromNode(pValueNode), false);
 
   colInfoDataEnsureCapacity(out->columnData, 1);
-  code = vectorConvertImpl(&in, out);
+  code = vectorConvertImpl(&in, out, overflow);
   sclFreeParam(&in);
 
   return code;
@@ -107,15 +107,21 @@ int32_t scalarGenerateSetFromList(void **data, void *pNode, uint32_t type) {
         out.columnData->info.bytes = tDataTypes[type].bytes;
       }
 
-      code = doConvertDataType(valueNode, &out);
+      int32_t overflow = 0;
+      code = doConvertDataType(valueNode, &out, &overflow);
       if (code != TSDB_CODE_SUCCESS) {
 //        sclError("convert data from %d to %d failed", in.type, out.type);
         SCL_ERR_JRET(code);
       }
 
+      if (overflow) {
+        cell = cell->pNext;
+        continue;
+      }
+
       if (IS_VAR_DATA_TYPE(type)) {
         buf = colDataGetVarData(out.columnData, 0);
-        len = varDataTLen(data);
+        len = varDataTLen(buf);
       } else {
         len = tDataTypes[type].bytes;
         buf = out.columnData->pData;

source/libs/scalar/src/sclvector.c

@@ -276,7 +276,7 @@ _getValueAddr_fn_t getVectorValueAddrFn(int32_t srcType) {
     return p;
 }
 
-static FORCE_INLINE void varToTimestamp(char *buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToTimestamp(char *buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
   int64_t value = 0;
   if (taosParseTime(buf, &value, strlen(buf), pOut->columnData->info.precision, tsDaylight) != TSDB_CODE_SUCCESS) {
     value = 0;
@@ -285,10 +285,26 @@ static FORCE_INLINE void varToTimestamp(char *buf, SScalarParam* pOut, int32_t r
   colDataAppendInt64(pOut->columnData, rowIndex, &value);
 }
 
-static FORCE_INLINE void varToSigned(char *buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToSigned(char *buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
+  if (overflow) {
+    int64_t minValue = tDataTypes[pOut->columnData->info.type].minValue;
+    int64_t maxValue = tDataTypes[pOut->columnData->info.type].maxValue;
+    int64_t value = (int64_t)taosStr2Int64(buf, NULL, 10);
+    if (value > maxValue) {
+      *overflow = 1;
+      return;
+    } else if (value < minValue) {
+      *overflow = -1;
+      return;
+    } else {
+      *overflow = 0;
+    }
+  }
+      
   switch (pOut->columnData->info.type) {
     case TSDB_DATA_TYPE_TINYINT: {
       int8_t value = (int8_t)taosStr2Int8(buf, NULL, 10);
+     
       colDataAppendInt8(pOut->columnData, rowIndex, (int8_t*)&value);
       break;
     } 
@@ -310,7 +326,22 @@ static FORCE_INLINE void varToSigned(char *buf, SScalarParam* pOut, int32_t rowI
   }
 }
 
-static FORCE_INLINE void varToUnsigned(char *buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToUnsigned(char *buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
+  if (overflow) {
+    uint64_t minValue = (uint64_t)tDataTypes[pOut->columnData->info.type].minValue;
+    uint64_t maxValue = (uint64_t)tDataTypes[pOut->columnData->info.type].maxValue;
+    uint64_t value = (uint64_t)taosStr2UInt64(buf, NULL, 10);
+    if (value > maxValue) {
+      *overflow = 1;
+      return;
+    } else if (value < minValue) {
+      *overflow = -1;
+      return;
+    } else {
+      *overflow = 0;
+    }
+  }
+
   switch (pOut->columnData->info.type) {
     case TSDB_DATA_TYPE_UTINYINT: {
       uint8_t value = (uint8_t)taosStr2UInt8(buf, NULL, 10);
@@ -335,18 +366,24 @@ static FORCE_INLINE void varToUnsigned(char *buf, SScalarParam* pOut, int32_t ro
   }
 }
 
-static FORCE_INLINE void varToFloat(char *buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToFloat(char *buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
+  if (TSDB_DATA_TYPE_FLOAT == pOut->columnData->info.type) {
+    float value = taosStr2Float(buf, NULL);
+    colDataAppendFloat(pOut->columnData, rowIndex, &value);
+    return;
+  }
+  
   double value = taosStr2Double(buf, NULL);
   colDataAppendDouble(pOut->columnData, rowIndex, &value);
 }
 
-static FORCE_INLINE void varToBool(char *buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToBool(char *buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
   int64_t value = taosStr2Int64(buf, NULL, 10);
   bool v = (value != 0)? true:false;
   colDataAppendInt8(pOut->columnData, rowIndex, (int8_t*) &v);
 }
 
-static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
   int32_t len = 0;
   int32_t inputLen = varDataLen(buf);
   int32_t outputMaxLen = (inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE;
@@ -359,7 +396,7 @@ static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIn
   taosMemoryFree(t);
 }
 
-static FORCE_INLINE void ncharToVar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
+static FORCE_INLINE void ncharToVar(char* buf, SScalarParam* pOut, int32_t rowIndex, int32_t* overflow) {
   int32_t inputLen = varDataLen(buf);
 
   char* t = taosMemoryCalloc(1, inputLen + VARSTR_HEADER_SIZE);
@@ -376,7 +413,7 @@ static FORCE_INLINE void ncharToVar(char* buf, SScalarParam* pOut, int32_t rowIn
 
 
 //TODO opt performance, tmp is not needed.
-int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
+int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType, int32_t* overflow) {
   bool vton = false;
 
   _bufConverteFunc func = NULL;
@@ -415,8 +452,7 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
     if(inType == TSDB_DATA_TYPE_JSON){
       if(*data == TSDB_DATA_TYPE_NULL) {
         ASSERT(0);
-      }
-      else if(*data == TSDB_DATA_TYPE_NCHAR) {
+      } else if(*data == TSDB_DATA_TYPE_NCHAR) {
         data += CHAR_BYTES;
         convertType = TSDB_DATA_TYPE_NCHAR;
       } else if(tTagIsJson(data)){
@@ -453,7 +489,7 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
       }
     }
     
-    (*func)(tmp, pOut, i);
+    (*func)(tmp, pOut, i, overflow);
     taosMemoryFreeClear(tmp);
   }
   
@@ -606,7 +642,7 @@ int32_t vectorConvertToVarData(const SScalarParam* pIn, SScalarParam* pOut, int1
       int32_t len = sprintf(varDataVal(tmp), "%" PRId64, value);
       varDataLen(tmp) = len;
       if (outType == TSDB_DATA_TYPE_NCHAR) {
-        varToNchar(tmp, pOut, i);
+        varToNchar(tmp, pOut, i, NULL);
       } else {
         colDataAppend(pOutputCol, i, (char *)tmp, false);
       }
@@ -623,7 +659,7 @@ int32_t vectorConvertToVarData(const SScalarParam* pIn, SScalarParam* pOut, int1
       int32_t len = sprintf(varDataVal(tmp), "%" PRIu64, value);
       varDataLen(tmp) = len;
       if (outType == TSDB_DATA_TYPE_NCHAR) {
-        varToNchar(tmp, pOut, i);
+        varToNchar(tmp, pOut, i, NULL);
       } else {
         colDataAppend(pOutputCol, i, (char *)tmp, false);
       }
@@ -640,7 +676,7 @@ int32_t vectorConvertToVarData(const SScalarParam* pIn, SScalarParam* pOut, int1
       int32_t len = sprintf(varDataVal(tmp), "%lf", value);
       varDataLen(tmp) = len;
       if (outType == TSDB_DATA_TYPE_NCHAR) {
-        varToNchar(tmp, pOut, i);
+        varToNchar(tmp, pOut, i, NULL);
       } else {
         colDataAppend(pOutputCol, i, (char *)tmp, false);
       }
@@ -653,9 +689,8 @@ int32_t vectorConvertToVarData(const SScalarParam* pIn, SScalarParam* pOut, int1
   return TSDB_CODE_SUCCESS;
 }
 
-
 // TODO opt performance
-int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut) {
+int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut, int32_t* overflow) {
   SColumnInfoData* pInputCol  = pIn->columnData;
   SColumnInfoData* pOutputCol = pOut->columnData;
 
@@ -668,7 +703,47 @@ int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut) {
   int16_t outType  = pOutputCol->info.type;
 
   if (IS_VAR_DATA_TYPE(inType)) {
-    return vectorConvertFromVarData(pIn, pOut, inType, outType);
+    return vectorConvertFromVarData(pIn, pOut, inType, outType, overflow);
+  }
+
+  if (overflow) {
+    ASSERT(1 == pIn->numOfRows);
+
+    pOut->numOfRows = 0;
+    
+    if (IS_SIGNED_NUMERIC_TYPE(outType)) {
+      int64_t minValue = tDataTypes[outType].minValue;
+      int64_t maxValue = tDataTypes[outType].maxValue;
+      
+      double value = 0;
+      GET_TYPED_DATA(value, double, inType, colDataGetData(pInputCol, 0));
+      
+      if (value > maxValue) {
+        *overflow = 1;
+        return TSDB_CODE_SUCCESS;
+      } else if (value < minValue) {
+        *overflow = -1;
+        return TSDB_CODE_SUCCESS;
+      } else {
+        *overflow = 0;
+      }
+    } else if (IS_UNSIGNED_NUMERIC_TYPE(outType)) {
+      uint64_t minValue = (uint64_t)tDataTypes[outType].minValue;
+      uint64_t maxValue = (uint64_t)tDataTypes[outType].maxValue;
+      
+      double value = 0;
+      GET_TYPED_DATA(value, double, inType, colDataGetData(pInputCol, 0));
+      
+      if (value > maxValue) {
+        *overflow = 1;
+        return TSDB_CODE_SUCCESS;
+      } else if (value < minValue) {
+        *overflow = -1;
+        return TSDB_CODE_SUCCESS;
+      } else {
+        *overflow = 0;
+      }
+    }
   }
 
   pOut->numOfRows = pIn->numOfRows;
@@ -829,6 +904,8 @@ int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut) {
   return TSDB_CODE_SUCCESS;
 }
 
+
+
 int8_t gConvertTypes[TSDB_DATA_TYPE_BLOB+1][TSDB_DATA_TYPE_BLOB+1] = {
 /*         NULL BOOL TINY SMAL INT  BIG  FLOA DOUB VARC TIME NCHA UTIN USMA UINT UBIG JSON VARB DECI BLOB */
 /*NULL*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
@@ -873,7 +950,7 @@ int32_t vectorConvertScalarParam(SScalarParam *input, SScalarParam *output, int3
     return TSDB_CODE_OUT_OF_MEMORY;
   }
 
-  code = vectorConvertImpl(input, output);
+  code = vectorConvertImpl(input, output, NULL);
   if (code) {
 //      taosMemoryFreeClear(paramOut1->data);
     return code;
@@ -944,7 +1021,7 @@ static int32_t doConvertHelper(SScalarParam* pDest, int32_t* convert, const SSca
       return code;
     }
 
-    code = vectorConvertImpl(pParam, pDest);
+    code = vectorConvertImpl(pParam, pDest, NULL);
     if (code != TSDB_CODE_SUCCESS) {
       return code;
     }
@@ -1714,7 +1791,7 @@ void vectorNotNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
 }
 
 void vectorIsTrue(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
-  vectorConvertImpl(pLeft, pOut);
+  vectorConvertImpl(pLeft, pOut, NULL);
   for(int32_t i = 0; i < pOut->numOfRows; ++i) {
     if(colDataIsNull_s(pOut->columnData, i)) {
       int8_t v = 0;

tests/script/jenkins/basic.txt

@@ -52,6 +52,7 @@
 ./test.sh -f tsim/insert/null.sim
 ./test.sh -f tsim/insert/update0.sim
 ./test.sh -f tsim/insert/commit-merge0.sim
+./test.sh -f tsim/insert/insert_select.sim
 
 # ---- parser
 ./test.sh -f tsim/parser/groupby-basic.sim
@@ -193,4 +194,7 @@
 # --- catalog
 ./test.sh -f tsim/catalog/alterInCurrent.sim
 
+# --- scalar
+./test.sh -f tsim/scalar/in.sim
+
 #======================b1-end===============

tests/script/tsim/scalar/in.sim

+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/exec.sh -n dnode1 -s start
+sql connect
+
+print ======== step1 
+sql drop database if exists db1;
+sql create database db1 vgroups 3;
+sql use db1;
+sql create stable st1 (fts timestamp, fbool bool, ftiny tinyint, fsmall smallint, fint int, fbig bigint, futiny tinyint unsigned, fusmall smallint unsigned, fuint int unsigned, fubig bigint unsigned, ffloat float, fdouble double, fbin binary(10), fnchar nchar(10)) tags(tts timestamp, tbool bool, ttiny tinyint, tsmall smallint, tint int, tbig bigint, tutiny tinyint unsigned, tusmall smallint unsigned, tuint int unsigned, tubig bigint unsigned, tfloat float, tdouble double, tbin binary(10), tnchar nchar(10));
+sql create table tb1 using st1 tags('2022-07-10 16:31:00', true, 1, 1, 1, 1, 1, 1, 1, 1, 1.0, 1.0, 'a', 'a');
+sql create table tb2 using st1 tags('2022-07-10 16:32:00', false, 2, 2, 2, 2, 2, 2, 2, 2, 2.0, 2.0, 'b', 'b');
+sql create table tb3 using st1 tags('2022-07-10 16:33:00', true, 3, 3, 3, 3, 3, 3, 3, 3, 3.0, 3.0, 'c', 'c');
+ 
+sql insert into tb1 values ('2022-07-10 16:31:01', false, 1, 1, 1, 1, 1, 1, 1, 1, 1.0, 1.0, 'a', 'a');
+sql insert into tb1 values ('2022-07-10 16:31:02', true, 2, 2, 2, 2, 2, 2, 2, 2, 2.0, 2.0, 'b', 'b');
+sql insert into tb1 values ('2022-07-10 16:31:03', false, 3, 3, 3, 3, 3, 3, 3, 3, 3.0, 3.0, 'c', 'c');
+sql insert into tb1 values ('2022-07-10 16:31:04', true, 4, 4, 4, 4, 4, 4, 4, 4, 4.0, 4.0, 'd', 'd');
+sql insert into tb1 values ('2022-07-10 16:31:05', false, 5, 5, 5, 5, 5, 5, 5, 5, 5.0, 5.0, 'e', 'e');
+ 
+sql insert into tb2 values ('2022-07-10 16:32:01', false, 1, 1, 1, 1, 1, 1, 1, 1, 1.0, 1.0, 'a', 'a');
+sql insert into tb2 values ('2022-07-10 16:32:02', true, 2, 2, 2, 2, 2, 2, 2, 2, 2.0, 2.0, 'b', 'b'); 
+sql insert into tb2 values ('2022-07-10 16:32:03', false, 3, 3, 3, 3, 3, 3, 3, 3, 3.0, 3.0, 'c', 'c');
+sql insert into tb2 values ('2022-07-10 16:32:04', true, 4, 4, 4, 4, 4, 4, 4, 4, 4.0, 4.0, 'd', 'd'); 
+sql insert into tb2 values ('2022-07-10 16:32:05', false, 5, 5, 5, 5, 5, 5, 5, 5, 5.0, 5.0, 'e', 'e');
+ 
+sql insert into tb3 values ('2022-07-10 16:33:01', false, 1, 1, 1, 1, 1, 1, 1, 1, 1.0, 1.0, 'a', 'a');
+sql insert into tb3 values ('2022-07-10 16:33:02', true, 2, 2, 2, 2, 2, 2, 2, 2, 2.0, 2.0, 'b', 'b'); 
+sql insert into tb3 values ('2022-07-10 16:33:03', false, 3, 3, 3, 3, 3, 3, 3, 3, 3.0, 3.0, 'c', 'c');
+sql insert into tb3 values ('2022-07-10 16:33:04', true, 4, 4, 4, 4, 4, 4, 4, 4, 4.0, 4.0, 'd', 'd'); 
+sql insert into tb3 values ('2022-07-10 16:33:05', false, 5, 5, 5, 5, 5, 5, 5, 5, 5.0, 5.0, 'e', 'e'); 
+
+sql select * from tb1 where fts in ('2022-07-10 16:31:01', '2022-07-10 16:31:03', 1657441865000);
+if $rows != 3 then
+  return -1
+endi
+
+sql select * from tb1 where fbool in (0, 3);
+if $rows != 5 then
+  return -1
+endi
+
+sql select * from tb1 where ftiny in (257);
+if $rows != 0 then
+  return -1
+endi
+
+sql select * from tb1 where ftiny in (2, 257);
+if $rows != 1 then
+  return -1
+endi
+
+sql select * from tb1 where futiny in (0, 257);
+if $rows != 0 then
+  return -1
+endi
+
+sql select * from st1 where tts in ('2022-07-10 16:31:00', '2022-07-10 16:33:00', 1657441865000);
+if $rows != 10 then
+  return -1
+endi
+
+sql select * from st1 where tbool in (0, 3);
+if $rows != 15 then
+  return -1
+endi
+
+sql select * from st1 where ttiny in (257);
+if $rows != 0 then
+  return -1
+endi
+
+sql select * from st1 where ttiny in (2, 257);
+if $rows != 5 then
+  return -1
+endi
+
+sql select * from st1 where tutiny in (0, 257);
+if $rows != 0 then
+  return -1
+endi
+
+system sh/exec.sh -n dnode1 -s stop -x SIGINT