Merge pull request #12274 from taosdata/feature/3.0_liaohj

enh(query):support selectivity function and normal column data query.

source/libs/executor/src/executorimpl.c

@@ -13,6 +13,7 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include <libs/function/function.h>
 #include "filter.h"
 #include "function.h"
 #include "functionMgt.h"
@@ -803,7 +804,8 @@ static void doAggregateImpl(SOperatorInfo* pOperator, TSKEY startTs, SqlFunction
   for (int32_t k = 0; k < pOperator->numOfExprs; ++k) {
     if (functionNeedToExecute(&pCtx[k])) {
       pCtx[k].startTs = startTs;  // this can be set during create the struct
-      pCtx[k].fpSet.process(&pCtx[k]);
+      if (pCtx[k].fpSet.process != NULL)
+        pCtx[k].fpSet.process(&pCtx[k]);
     }
   }
 }
@@ -1074,35 +1076,36 @@ void setBlockStatisInfo(SqlFunctionCtx* pCtx, SExprInfo* pExprInfo, SSDataBlock*
 // set the output buffer for the selectivity + tag query
 static int32_t setCtxTagColumnInfo(SqlFunctionCtx* pCtx, int32_t numOfOutput) {
   int32_t num = 0;
-  int16_t tagLen = 0;
 
   SqlFunctionCtx*  p = NULL;
-  SqlFunctionCtx** pTagCtx = taosMemoryCalloc(numOfOutput, POINTER_BYTES);
-  if (pTagCtx == NULL) {
+  SqlFunctionCtx** pValCtx = taosMemoryCalloc(numOfOutput, POINTER_BYTES);
+  if (pValCtx == NULL) {
     return TSDB_CODE_QRY_OUT_OF_MEMORY;
   }
 
   for (int32_t i = 0; i < numOfOutput; ++i) {
-    int32_t functionId = pCtx[i].functionId;
-
-    if (functionId == FUNCTION_TAG_DUMMY || functionId == FUNCTION_TS_DUMMY) {
-      tagLen += pCtx[i].resDataInfo.bytes;
-      pTagCtx[num++] = &pCtx[i];
-    } else if (1 /*(aAggs[functionId].status & FUNCSTATE_SELECTIVITY) != 0*/) {
-      p = &pCtx[i];
-    } else if (functionId == FUNCTION_TS || functionId == FUNCTION_TAG) {
-      // tag function may be the group by tag column
-      // ts may be the required primary timestamp column
-      continue;
+    if (strcmp(pCtx[i].pExpr->pExpr->_function.functionName, "_select_value") == 0) {
+      pValCtx[num++] = &pCtx[i];
     } else {
-      // the column may be the normal column, group by normal_column, the functionId is FUNCTION_PRJ
+      p = &pCtx[i];
     }
+//    if (functionId == FUNCTION_TAG_DUMMY || functionId == FUNCTION_TS_DUMMY) {
+//      tagLen += pCtx[i].resDataInfo.bytes;
+//      pTagCtx[num++] = &pCtx[i];
+//    } else if (functionId == FUNCTION_TS || functionId == FUNCTION_TAG) {
+//      // tag function may be the group by tag column
+//      // ts may be the required primary timestamp column
+//      continue;
+//    } else {
+//      // the column may be the normal column, group by normal_column, the functionId is FUNCTION_PRJ
+//    }
   }
+
   if (p != NULL) {
-    p->subsidiaries.pCtx = pTagCtx;
+    p->subsidiaries.pCtx = pValCtx;
     p->subsidiaries.num = num;
   } else {
-    taosMemoryFreeClear(pTagCtx);
+    taosMemoryFreeClear(pValCtx);
   }
 
   return TSDB_CODE_SUCCESS;
@@ -2219,6 +2222,8 @@ int32_t doCopyToSDataBlock(SSDataBlock* pBlock, SExprInfo* pExprInfo, SDiskbased
       pCtx[j].resultInfo = getResultCell(pRow, j, rowCellOffset);
       if (pCtx[j].fpSet.process) {
         pCtx[j].fpSet.finalize(&pCtx[j], pBlock);
+      } else if (strcmp(pCtx[j].pExpr->pExpr->_function.functionName, "_select_value") == 0) {
+        // do nothing, todo refactor
       } else {
         SColumnInfoData* pColInfoData = taosArrayGet(pBlock->pDataBlock, slotId);
 
@@ -3974,6 +3979,7 @@ static SSDataBlock* doProjectOperation(SOperatorInfo* pOperator) {
   SSDataBlock* pRes = pInfo->pRes;
   blockDataCleanup(pRes);
 
+  SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
   if (pOperator->status == OP_EXEC_DONE) {
     return NULL;
   }
@@ -4037,9 +4043,13 @@ static SSDataBlock* doProjectOperation(SOperatorInfo* pOperator) {
     setInputDataBlock(pOperator, pInfo->pCtx, pBlock, order, false);
     blockDataEnsureCapacity(pInfo->pRes, pInfo->pRes->info.rows + pBlock->info.rows);
 
-    projectApplyFunctions(pOperator->pExpr, pInfo->pRes, pBlock, pInfo->pCtx, pOperator->numOfExprs,
+    pTaskInfo->code = projectApplyFunctions(pOperator->pExpr, pInfo->pRes, pBlock, pInfo->pCtx, pOperator->numOfExprs,
                           pProjectInfo->pPseudoColInfo);
 
+    if (pTaskInfo->code != TSDB_CODE_SUCCESS) {
+      longjmp(pTaskInfo->env, pTaskInfo->code);
+    }
+
     int32_t status = handleLimitOffset(pOperator, pBlock);
     if (status == PROJECT_RETRIEVE_CONTINUE) {
       continue;

source/libs/executor/src/groupoperator.c

@@ -262,6 +262,8 @@ static SSDataBlock* hashGroupbyAggregate(SOperatorInfo* pOperator) {
     return NULL;
   }
 
+  SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
+
   SGroupbyOperatorInfo* pInfo = pOperator->info;
   SSDataBlock* pRes = pInfo->binfo.pRes;
 
@@ -289,7 +291,10 @@ static SSDataBlock* hashGroupbyAggregate(SOperatorInfo* pOperator) {
 
     // there is an scalar expression that needs to be calculated right before apply the group aggregation.
     if (pInfo->pScalarExprInfo != NULL) {
-      projectApplyFunctions(pInfo->pScalarExprInfo, pBlock, pBlock, pInfo->pScalarFuncCtx, pInfo->numOfScalarExpr, NULL);
+      pTaskInfo->code = projectApplyFunctions(pInfo->pScalarExprInfo, pBlock, pBlock, pInfo->pScalarFuncCtx, pInfo->numOfScalarExpr, NULL);
+      if (pTaskInfo->code != TSDB_CODE_SUCCESS) {
+        longjmp(pTaskInfo->env, pTaskInfo->code);
+      }
     }
 
     //    setTagValue(pOperator, pRuntimeEnv->current->pTable, pInfo->binfo.pCtx, pOperator->numOfExprs);

source/libs/executor/src/sortoperator.c

@@ -114,7 +114,10 @@ void applyScalarFunction(SSDataBlock* pBlock, void* param) {
   SOperatorInfo* pOperator = param;
   SSortOperatorInfo* pSort = pOperator->info;
   if (pOperator->pExpr != NULL) {
-    projectApplyFunctions(pOperator->pExpr, pBlock, pBlock, pSort->binfo.pCtx, pOperator->numOfExprs, NULL);
+    int32_t code = projectApplyFunctions(pOperator->pExpr, pBlock, pBlock, pSort->binfo.pCtx, pOperator->numOfExprs, NULL);
+    if (code != TSDB_CODE_SUCCESS) {
+      longjmp(pOperator->pTaskInfo->env, code);
+    }
   }
 }
 

source/libs/function/inc/builtinsimpl.h

@@ -37,11 +37,11 @@ bool getSumFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 int32_t sumFunction(SqlFunctionCtx *pCtx);
 int32_t sumInvertFunction(SqlFunctionCtx *pCtx);
 
-bool minFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
-bool maxFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
+bool minmaxFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
 bool getMinmaxFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 int32_t minFunction(SqlFunctionCtx* pCtx);
 int32_t maxFunction(SqlFunctionCtx *pCtx);
+int32_t minmaxFunctionFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
 
 bool getAvgFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 bool avgFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
@@ -70,6 +70,7 @@ int32_t lastFunction(SqlFunctionCtx *pCtx);
 
 bool getTopBotFuncEnv(SFunctionNode* UNUSED_PARAM(pFunc), SFuncExecEnv* pEnv);
 int32_t topFunction(SqlFunctionCtx *pCtx);
+int32_t bottomFunction(SqlFunctionCtx *pCtx);
 int32_t topBotFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
 
 bool getSpreadFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
@@ -82,6 +83,8 @@ bool histogramFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultIn
 int32_t histogramFunction(SqlFunctionCtx* pCtx);
 int32_t histogramFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
 
+bool getSelectivityFuncEnv(SFunctionNode* pFunc, SFuncExecEnv* pEnv);
+
 #ifdef __cplusplus
 }
 #endif

source/libs/function/src/builtins.c

@@ -207,7 +207,8 @@ static int32_t translateTop(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
 }
 
 static int32_t translateBottom(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
-  // todo
+  SDataType* pType = &((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType;
+  pFunc->node.resType = (SDataType){.bytes = pType->bytes, .type = pType->type};
   return TSDB_CODE_SUCCESS;
 }
 
@@ -509,9 +510,9 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .translateFunc = translateInOutNum,
     .dataRequiredFunc = statisDataRequired,
     .getEnvFunc   = getMinmaxFuncEnv,
-    .initFunc     = minFunctionSetup,
+    .initFunc     = minmaxFunctionSetup,
     .processFunc  = minFunction,
-    .finalizeFunc = functionFinalize
+    .finalizeFunc = minmaxFunctionFinalize
   },
   {
     .name = "max",
@@ -520,9 +521,9 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .translateFunc = translateInOutNum,
     .dataRequiredFunc = statisDataRequired,
     .getEnvFunc   = getMinmaxFuncEnv,
-    .initFunc     = maxFunctionSetup,
+    .initFunc     = minmaxFunctionSetup,
     .processFunc  = maxFunction,
-    .finalizeFunc = functionFinalize
+    .finalizeFunc = minmaxFunctionFinalize
   },
   {
     .name = "stddev",
@@ -562,14 +563,14 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .classification = FUNC_MGT_AGG_FUNC,
     .translateFunc = translateApercentile,
     .getEnvFunc   = getMinmaxFuncEnv,
-    .initFunc     = maxFunctionSetup,
+    .initFunc     = minmaxFunctionSetup,
     .processFunc  = maxFunction,
     .finalizeFunc = functionFinalize
   },
   {
     .name = "top",
     .type = FUNCTION_TYPE_TOP,
-    .classification = FUNC_MGT_AGG_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC,
     .translateFunc = translateTop,
     .getEnvFunc   = getTopBotFuncEnv,
     .initFunc     = functionSetup,
@@ -579,12 +580,12 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "bottom",
     .type = FUNCTION_TYPE_BOTTOM,
-    .classification = FUNC_MGT_AGG_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC,
     .translateFunc = translateBottom,
-    .getEnvFunc   = getMinmaxFuncEnv,
-    .initFunc     = maxFunctionSetup,
-    .processFunc  = maxFunction,
-    .finalizeFunc = functionFinalize
+    .getEnvFunc   = getTopBotFuncEnv,
+    .initFunc     = functionSetup,
+    .processFunc  = bottomFunction,
+    .finalizeFunc = topBotFinalize
   },
   {
     .name = "spread",
@@ -603,7 +604,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_MULTI_RES_FUNC,
     .translateFunc = translateLastRow,
     .getEnvFunc   = getMinmaxFuncEnv,
-    .initFunc     = maxFunctionSetup,
+    .initFunc     = minmaxFunctionSetup,
     .processFunc  = maxFunction,
     .finalizeFunc = functionFinalize
   },
@@ -1032,8 +1033,8 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .type = FUNCTION_TYPE_SELECT_VALUE,
     .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC,
     .translateFunc = translateSelectValue,
-    .getEnvFunc   = NULL,
-    .initFunc     = NULL,
+    .getEnvFunc   = getSelectivityFuncEnv,  // todo remove this function later.
+    .initFunc     = functionSetup,
     .sprocessFunc = NULL,
     .finalizeFunc = NULL
   }

tests/script/tsim/insert/basic0.sim

@@ -140,7 +140,8 @@ endi
 if $data00 != -13 then 
   return -1
 endi
-if $data01 != -2.30000 then 
+if $data01 != -2.30000 then
+  print expect -2.30000, actual: $data01
   return -1
 endi
 if $data02 != -3.300000000 then