xxq250
/
homework-jianmu
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge pull request #12614 from taosdata/feature/3.0_liaohj 

fix(query): expand the capacity for aggregate operator.
This commit is contained in:
Haojun Liao 2022-05-18 09:56:15 +08:00 committed by GitHub
parent 6e1d926f36 52f6a66ba5
commit ee61621b9e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 40 additions and 697 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
include/libs/function/function.h
View File

@ -170,32 +170,18 @@ typedef struct SInputColumnInfoData {
// sql function runtime context // sql function runtime context
typedef struct SqlFunctionCtx { typedef struct SqlFunctionCtx {
SInputColumnInfoData input; SInputColumnInfoData input;
SResultDataInfo resDataInfo; SResultDataInfo resDataInfo;
uint32_t order; // data block scanner order: asc|desc uint32_t order; // data block scanner order: asc|desc
uint8_t scanFlag; // record current running step, default: 0 uint8_t scanFlag; // record current running step, default: 0
//////////////////////////////////////////////////////////////// int16_t functionId; // function id
int32_t startRow; // start row index char *pOutput; // final result output buffer, point to sdata->data
int32_t size; // handled processed row number int32_t numOfParams;
SColumnInfoData* pInput; SFunctParam *param; // input parameter, e.g., top(k, 20), the number of results for top query is kept in param
SColumnDataAgg agg; int64_t *ptsList; // corresponding timestamp array list
int16_t inputType; // TODO remove it SColumnInfoData *pTsOutput; // corresponding output buffer for timestamp of each result, e.g., top/bottom*/
int16_t inputBytes; // TODO remove it int32_t offset;
bool hasNull; // null value exist in current block, TODO remove it SVariant tag;
bool requireNull; // require null in some function, TODO remove it
int32_t columnIndex; // TODO remove it
bool isAggSet;
int64_t startTs; // timestamp range of current query when function is executed on a specific data block, TODO remove it
bool stableQuery;
/////////////////////////////////////////////////////////////////
int16_t functionId; // function id
char * pOutput; // final result output buffer, point to sdata->data
int32_t numOfParams;
SFunctParam *param; // input parameter, e.g., top(k, 20), the number of results for top query is kept in param
int64_t *ptsList; // corresponding timestamp array list
SColumnInfoData *pTsOutput; // corresponding output buffer for timestamp of each result, e.g., top/bottom*/
int32_t offset;
SVariant tag;
struct SResultRowEntryInfo *resultInfo; struct SResultRowEntryInfo *resultInfo;
SSubsidiaryResInfo subsidiaries; SSubsidiaryResInfo subsidiaries;
SPoint1 start; SPoint1 start;

18
source/libs/executor/src/executorimpl.c
View File

@ -602,8 +602,6 @@ void initExecTimeWindowInfo(SColumnInfoData* pColData, STimeWindow* pQueryWindow
void doApplyFunctions(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, STimeWindow* pWin, SColumnInfoData* pTimeWindowData, int32_t offset, void doApplyFunctions(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, STimeWindow* pWin, SColumnInfoData* pTimeWindowData, int32_t offset,
int32_t forwardStep, TSKEY* tsCol, int32_t numOfTotal, int32_t numOfOutput, int32_t order) { int32_t forwardStep, TSKEY* tsCol, int32_t numOfTotal, int32_t numOfOutput, int32_t order) {
for (int32_t k = 0; k < numOfOutput; ++k) { for (int32_t k = 0; k < numOfOutput; ++k) {
pCtx[k].startTs = pWin->skey;
// keep it temporarily // keep it temporarily
bool hasAgg = pCtx[k].input.colDataAggIsSet; bool hasAgg = pCtx[k].input.colDataAggIsSet;
int32_t numOfRows = pCtx[k].input.numOfRows; int32_t numOfRows = pCtx[k].input.numOfRows;
@ -619,8 +617,8 @@ void doApplyFunctions(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, STimeWindow
// not a whole block involved in query processing, statistics data can not be used // not a whole block involved in query processing, statistics data can not be used
// NOTE: the original value of isSet have been changed here // NOTE: the original value of isSet have been changed here
if (pCtx[k].isAggSet && forwardStep < numOfTotal) { if (pCtx[k].input.colDataAggIsSet && forwardStep < numOfTotal) {
pCtx[k].isAggSet = false; pCtx[k].input.colDataAggIsSet = false;
} }
if (fmIsWindowPseudoColumnFunc(pCtx[k].functionId)) { if (fmIsWindowPseudoColumnFunc(pCtx[k].functionId)) {
@ -680,7 +678,7 @@ static void doSetInputDataBlockInfo(SOperatorInfo* pOperator, SqlFunctionCtx* pC
int32_t order) { int32_t order) {
for (int32_t i = 0; i < pOperator->numOfExprs; ++i) { for (int32_t i = 0; i < pOperator->numOfExprs; ++i) {
pCtx[i].order = order; pCtx[i].order = order;
pCtx[i].size = pBlock->info.rows; pCtx[i].input.numOfRows = pBlock->info.rows;
setBlockStatisInfo(&pCtx[i], &pOperator->pExpr[i], pBlock); setBlockStatisInfo(&pCtx[i], &pOperator->pExpr[i], pBlock);
} }
} }
@ -742,7 +740,8 @@ static int32_t doSetInputDataBlock(SOperatorInfo* pOperator, SqlFunctionCtx* pCt
for (int32_t i = 0; i < pOperator->numOfExprs; ++i) { for (int32_t i = 0; i < pOperator->numOfExprs; ++i) {
pCtx[i].order = order; pCtx[i].order = order;
pCtx[i].size = pBlock->info.rows; pCtx[i].input.numOfRows = pBlock->info.rows;
pCtx[i].pSrcBlock = pBlock; pCtx[i].pSrcBlock = pBlock;
pCtx[i].scanFlag = scanFlag; pCtx[i].scanFlag = scanFlag;
@ -827,7 +826,6 @@ static int32_t doSetInputDataBlock(SOperatorInfo* pOperator, SqlFunctionCtx* pCt
static int32_t doAggregateImpl(SOperatorInfo* pOperator, TSKEY startTs, SqlFunctionCtx* pCtx) { static int32_t doAggregateImpl(SOperatorInfo* pOperator, TSKEY startTs, SqlFunctionCtx* pCtx) {
for (int32_t k = 0; k < pOperator->numOfExprs; ++k) { for (int32_t k = 0; k < pOperator->numOfExprs; ++k) {
if (functionNeedToExecute(&pCtx[k])) { if (functionNeedToExecute(&pCtx[k])) {
pCtx[k].startTs = startTs;
// todo add a dummy funtion to avoid process check // todo add a dummy funtion to avoid process check
if (pCtx[k].fpSet.process != NULL) { if (pCtx[k].fpSet.process != NULL) {
int32_t code = pCtx[k].fpSet.process(&pCtx[k]); int32_t code = pCtx[k].fpSet.process(&pCtx[k]);
@ -3330,7 +3328,7 @@ static bool needToMerge(SSDataBlock* pBlock, SArray* groupInfo, char** buf, int3
static void doMergeResultImpl(SSortedMergeOperatorInfo* pInfo, SqlFunctionCtx* pCtx, int32_t numOfExpr, static void doMergeResultImpl(SSortedMergeOperatorInfo* pInfo, SqlFunctionCtx* pCtx, int32_t numOfExpr,
int32_t rowIndex) { int32_t rowIndex) {
for (int32_t j = 0; j < numOfExpr; ++j) { // TODO set row index for (int32_t j = 0; j < numOfExpr; ++j) { // TODO set row index
pCtx[j].startRow = rowIndex; // pCtx[j].startRow = rowIndex;
} }
for (int32_t j = 0; j < numOfExpr; ++j) { for (int32_t j = 0; j < numOfExpr; ++j) {
@ -3381,7 +3379,7 @@ static void doMergeImpl(SOperatorInfo* pOperator, int32_t numOfExpr, SSDataBlock
SqlFunctionCtx* pCtx = pInfo->binfo.pCtx; SqlFunctionCtx* pCtx = pInfo->binfo.pCtx;
for (int32_t i = 0; i < pBlock->info.numOfCols; ++i) { for (int32_t i = 0; i < pBlock->info.numOfCols; ++i) {
pCtx[i].size = 1; // pCtx[i].size = 1;
} }
for (int32_t i = 0; i < pBlock->info.rows; ++i) { for (int32_t i = 0; i < pBlock->info.rows; ++i) {
@ -4248,7 +4246,7 @@ SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SExprInfo*
goto _error; goto _error;
} }
int32_t numOfRows = 10; int32_t numOfRows = 1024;
size_t keyBufSize = sizeof(int64_t) + sizeof(int64_t) + POINTER_BYTES; size_t keyBufSize = sizeof(int64_t) + sizeof(int64_t) + POINTER_BYTES;
initResultSizeInfo(pOperator, numOfRows); initResultSizeInfo(pOperator, numOfRows);

29
source/libs/function/src/builtinsimpl.c
View File

@ -323,6 +323,7 @@ static FORCE_INLINE int32_t getNumofElem(SqlFunctionCtx* pCtx) {
} }
return numOfElem; return numOfElem;
} }
/* /*
* count function does need the finalize, if data is missing, the default value, which is 0, is used * count function does need the finalize, if data is missing, the default value, which is 0, is used
* count function does not use the pCtx->interResBuf to keep the intermediate buffer * count function does not use the pCtx->interResBuf to keep the intermediate buffer
@ -817,16 +818,6 @@ int32_t doMinMaxHelper(SqlFunctionCtx* pCtx, int32_t isMinFunc) {
int64_t val = GET_INT64_VAL(tval); int64_t val = GET_INT64_VAL(tval);
if ((prev < val) ^ isMinFunc) { if ((prev < val) ^ isMinFunc) {
pBuf->v = val; pBuf->v = val;
// for (int32_t i = 0; i < (pCtx)->subsidiaries.num; ++i) {
// SqlFunctionCtx* __ctx = pCtx->subsidiaries.pCtx[i];
// if (__ctx->functionId == FUNCTION_TS_DUMMY) { // TODO refactor
// __ctx->tag.i = key;
// __ctx->tag.nType = TSDB_DATA_TYPE_BIGINT;
// }
//
// __ctx->fpSet.process(__ctx);
// }
if (pCtx->subsidiaries.num > 0) { if (pCtx->subsidiaries.num > 0) {
saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos); saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos);
} }
@ -839,15 +830,6 @@ int32_t doMinMaxHelper(SqlFunctionCtx* pCtx, int32_t isMinFunc) {
uint64_t val = GET_UINT64_VAL(tval); uint64_t val = GET_UINT64_VAL(tval);
if ((prev < val) ^ isMinFunc) { if ((prev < val) ^ isMinFunc) {
pBuf->v = val; pBuf->v = val;
// for (int32_t i = 0; i < (pCtx)->subsidiaries.num; ++i) {
// SqlFunctionCtx* __ctx = pCtx->subsidiaries.pCtx[i];
// if (__ctx->functionId == FUNCTION_TS_DUMMY) { // TODO refactor
// __ctx->tag.i = key;
// __ctx->tag.nType = TSDB_DATA_TYPE_BIGINT;
// }
//
// __ctx->fpSet.process(__ctx);
// }
if (pCtx->subsidiaries.num > 0) { if (pCtx->subsidiaries.num > 0) {
saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos); saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos);
} }
@ -859,7 +841,6 @@ int32_t doMinMaxHelper(SqlFunctionCtx* pCtx, int32_t isMinFunc) {
double val = GET_DOUBLE_VAL(tval); double val = GET_DOUBLE_VAL(tval);
if ((prev < val) ^ isMinFunc) { if ((prev < val) ^ isMinFunc) {
pBuf->v = val; pBuf->v = val;
if (pCtx->subsidiaries.num > 0) { if (pCtx->subsidiaries.num > 0) {
saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos); saveTupleData(pCtx, index, pCtx->pSrcBlock, &pBuf->tuplePos);
} }
@ -1739,7 +1720,7 @@ int32_t percentileFunction(SqlFunctionCtx* pCtx) {
char* data = colDataGetData(pCol, i); char* data = colDataGetData(pCol, i);
double v = 0; double v = 0;
GET_TYPED_DATA(v, double, pCtx->inputType, data); GET_TYPED_DATA(v, double, type, data);
if (v < GET_DOUBLE_VAL(&pInfo->minval)) { if (v < GET_DOUBLE_VAL(&pInfo->minval)) {
SET_DOUBLE_VAL(&pInfo->minval, v); SET_DOUBLE_VAL(&pInfo->minval, v);
} }
@ -2552,7 +2533,7 @@ int32_t elapsedFunction(SqlFunctionCtx *pCtx) {
} }
} }
} else { // computing based on the true data block } else { // computing based on the true data block
if (0 == pCtx->size) { if (0 == pInput->numOfRows) {
if (pCtx->order == TSDB_ORDER_DESC) { if (pCtx->order == TSDB_ORDER_DESC) {
if (pCtx->end.key != INT64_MIN) { if (pCtx->end.key != INT64_MIN) {
pInfo->min = pCtx->end.key; pInfo->min = pCtx->end.key;
@ -2571,7 +2552,7 @@ int32_t elapsedFunction(SqlFunctionCtx *pCtx) {
TSKEY* ptsList = (int64_t*)colDataGetData(pCol, start); TSKEY* ptsList = (int64_t*)colDataGetData(pCol, start);
if (pCtx->order == TSDB_ORDER_DESC) { if (pCtx->order == TSDB_ORDER_DESC) {
if (pCtx->start.key == INT64_MIN) { if (pCtx->start.key == INT64_MIN) {
pInfo->max = (pInfo->max < ptsList[pCtx->size - 1]) ? ptsList[pCtx->size - 1] : pInfo->max; pInfo->max = (pInfo->max < ptsList[start + pInput->numOfRows - 1]) ? ptsList[start + pInput->numOfRows - 1] : pInfo->max;
} else { } else {
pInfo->max = pCtx->start.key + 1; pInfo->max = pCtx->start.key + 1;
} }
@ -2591,7 +2572,7 @@ int32_t elapsedFunction(SqlFunctionCtx *pCtx) {
if (pCtx->end.key != INT64_MIN) { if (pCtx->end.key != INT64_MIN) {
pInfo->max = pCtx->end.key + 1; pInfo->max = pCtx->end.key + 1;
} else { } else {
pInfo->max = ptsList[pCtx->size - 1]; pInfo->max = ptsList[start + pInput->numOfRows - 1];
} }
} }
} }

652
source/libs/function/src/taggfunction.c
View File

@ -480,7 +480,7 @@ static bool function_setup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInf
initResultRowEntry(pResultInfo, pCtx->resDataInfo.interBufSize); initResultRowEntry(pResultInfo, pCtx->resDataInfo.interBufSize);
return true; return true;
} }
#if 0
/** /**
* in handling the stable query, function_finalizer is called after the secondary * in handling the stable query, function_finalizer is called after the secondary
* merge being completed, during the first merge procedure, which is executed at the * merge being completed, during the first merge procedure, which is executed at the
@ -497,53 +497,6 @@ static void function_finalizer(SqlFunctionCtx *pCtx) {
doFinalizer(pCtx); doFinalizer(pCtx);
} }
/*
* count function does need the finalize, if data is missing, the default value, which is 0, is used
* count function does not use the pCtx->interResBuf to keep the intermediate buffer
*/
static void count_function(SqlFunctionCtx *pCtx) {
int32_t numOfElem = 0;
/*
* 1. column data missing (schema modified) causes pCtx->hasNull == true. pCtx->isAggSet == true;
* 2. for general non-primary key columns, pCtx->hasNull may be true or false, pCtx->isAggSet == true;
* 3. for primary key column, pCtx->hasNull always be false, pCtx->isAggSet == false;
*/
if (pCtx->isAggSet) {
numOfElem = pCtx->size - pCtx->agg.numOfNull;
} else {
if (pCtx->hasNull) {
for (int32_t i = 0; i < pCtx->size; ++i) {
char *val = GET_INPUT_DATA(pCtx, i);
if (isNull(val, pCtx->inputType)) {
continue;
}
numOfElem += 1;
}
} else {
//when counting on the primary time stamp column and no statistics data is presented, use the size value directly.
numOfElem = pCtx->size;
}
}
if (numOfElem > 0) {
// GET_RES_INFO(pCtx)->hasResult = DATA_SET_FLAG;
}
*((int64_t *)pCtx->pOutput) += numOfElem;
SET_VAL(pCtx, numOfElem, 1);
}
static void count_func_merge(SqlFunctionCtx *pCtx) {
int64_t *pData = (int64_t *)GET_INPUT_DATA_LIST(pCtx);
for (int32_t i = 0; i < pCtx->size; ++i) {
*((int64_t *)pCtx->pOutput) += pData[i];
}
SET_VAL(pCtx, pCtx->size, 1);
}
/** /**
* 1. If the column value for filter exists, we need to load the SFields, which serves * 1. If the column value for filter exists, we need to load the SFields, which serves
* as the pre-filter to decide if the actual data block is required or not. * as the pre-filter to decide if the actual data block is required or not.
@ -633,113 +586,6 @@ int32_t noDataRequired(SqlFunctionCtx *pCtx, STimeWindow* w, int32_t colId) {
LOOPCHECK_N(*_data, _list, ctx, tsdbType, sign, notNullElems); \ LOOPCHECK_N(*_data, _list, ctx, tsdbType, sign, notNullElems); \
} while (0) } while (0)
static void do_sum(SqlFunctionCtx *pCtx) {
int32_t notNullElems = 0;
// Only the pre-computing information loaded and actual data does not loaded
if (pCtx->isAggSet) {
notNullElems = pCtx->size - pCtx->agg.numOfNull;
assert(pCtx->size >= pCtx->agg.numOfNull);
if (IS_SIGNED_NUMERIC_TYPE(pCtx->inputType)) {
int64_t *retVal = (int64_t *)pCtx->pOutput;
*retVal += pCtx->agg.sum;
} else if (IS_UNSIGNED_NUMERIC_TYPE(pCtx->inputType)) {
uint64_t *retVal = (uint64_t *)pCtx->pOutput;
*retVal += (uint64_t)pCtx->agg.sum;
} else if (IS_FLOAT_TYPE(pCtx->inputType)) {
double *retVal = (double*) pCtx->pOutput;
SET_DOUBLE_VAL(retVal, *retVal + GET_DOUBLE_VAL((const char*)&(pCtx->agg.sum)));
}
} else { // computing based on the true data block
void *pData = GET_INPUT_DATA_LIST(pCtx);
notNullElems = 0;
if (IS_SIGNED_NUMERIC_TYPE(pCtx->inputType)) {
int64_t *retVal = (int64_t *)pCtx->pOutput;
if (pCtx->inputType == TSDB_DATA_TYPE_TINYINT) {
LIST_ADD_N(*retVal, pCtx, pData, int8_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_SMALLINT) {
LIST_ADD_N(*retVal, pCtx, pData, int16_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_INT) {
LIST_ADD_N(*retVal, pCtx, pData, int32_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_BIGINT) {
LIST_ADD_N(*retVal, pCtx, pData, int64_t, notNullElems, pCtx->inputType);
}
} else if (IS_UNSIGNED_NUMERIC_TYPE(pCtx->inputType)) {
uint64_t *retVal = (uint64_t *)pCtx->pOutput;
if (pCtx->inputType == TSDB_DATA_TYPE_UTINYINT) {
LIST_ADD_N(*retVal, pCtx, pData, uint8_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_USMALLINT) {
LIST_ADD_N(*retVal, pCtx, pData, uint16_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_UINT) {
LIST_ADD_N(*retVal, pCtx, pData, uint32_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_UBIGINT) {
LIST_ADD_N(*retVal, pCtx, pData, uint64_t, notNullElems, pCtx->inputType);
}
} else if (pCtx->inputType == TSDB_DATA_TYPE_DOUBLE) {
double *retVal = (double *)pCtx->pOutput;
LIST_ADD_N_DOUBLE(*retVal, pCtx, pData, double, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_FLOAT) {
double *retVal = (double *)pCtx->pOutput;
LIST_ADD_N_DOUBLE_FLOAT(*retVal, pCtx, pData, float, notNullElems, pCtx->inputType);
}
}
// data in the check operation are all null, not output
SET_VAL(pCtx, notNullElems, 1);
if (notNullElems > 0) {
// GET_RES_INFO(pCtx)->hasResult = DATA_SET_FLAG;
}
}
static void sum_function(SqlFunctionCtx *pCtx) {
do_sum(pCtx);
// keep the result data in output buffer, not in the intermediate buffer
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
// if (pResInfo->hasResult == DATA_SET_FLAG && pCtx->stableQuery) {
// set the flag for super table query
SSumInfo *pSum = (SSumInfo *)pCtx->pOutput;
pSum->hasResult = DATA_SET_FLAG;
// }
}
static void sum_func_merge(SqlFunctionCtx *pCtx) {
int32_t notNullElems = 0;
GET_TRUE_DATA_TYPE();
assert(pCtx->stableQuery);
for (int32_t i = 0; i < pCtx->size; ++i) {
char * input = GET_INPUT_DATA(pCtx, i);
SSumInfo *pInput = (SSumInfo *)input;
if (pInput->hasResult != DATA_SET_FLAG) {
continue;
}
notNullElems++;
if (IS_SIGNED_NUMERIC_TYPE(type)) {
*(int64_t *)pCtx->pOutput += pInput->isum;
} else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
*(uint64_t *) pCtx->pOutput += pInput->usum;
} else {
SET_DOUBLE_VAL((double *)pCtx->pOutput, *(double *)pCtx->pOutput + pInput->dsum);
}
}
SET_VAL(pCtx, notNullElems, 1);
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
if (notNullElems > 0) {
//pResInfo->hasResult = DATA_SET_FLAG;
}
}
static int32_t statisRequired(SqlFunctionCtx *pCtx, STimeWindow* w, int32_t colId) { static int32_t statisRequired(SqlFunctionCtx *pCtx, STimeWindow* w, int32_t colId) {
return BLK_DATA_SMA_LOAD; return BLK_DATA_SMA_LOAD;
} }
@ -822,17 +668,17 @@ static int32_t lastDistFuncRequired(SqlFunctionCtx *pCtx, STimeWindow* w, int32_
*/ */
static void avg_function(SqlFunctionCtx *pCtx) { static void avg_function(SqlFunctionCtx *pCtx) {
int32_t notNullElems = 0; int32_t notNullElems = 0;
// NOTE: keep the intermediate result into the interResultBuf // NOTE: keep the intermediate result into the interResultBuf
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx); SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
SAvgInfo *pAvgInfo = (SAvgInfo *)GET_ROWCELL_INTERBUF(pResInfo); SAvgInfo *pAvgInfo = (SAvgInfo *)GET_ROWCELL_INTERBUF(pResInfo);
double *pVal = &pAvgInfo->sum; double *pVal = &pAvgInfo->sum;
if (pCtx->isAggSet) { // Pre-aggregation if (pCtx->isAggSet) { // Pre-aggregation
notNullElems = pCtx->size - pCtx->agg.numOfNull; notNullElems = pCtx->size - pCtx->agg.numOfNull;
assert(notNullElems >= 0); assert(notNullElems >= 0);
if (IS_SIGNED_NUMERIC_TYPE(pCtx->inputType)) { if (IS_SIGNED_NUMERIC_TYPE(pCtx->inputType)) {
*pVal += pCtx->agg.sum; *pVal += pCtx->agg.sum;
} else if (IS_UNSIGNED_NUMERIC_TYPE(pCtx->inputType)) { } else if (IS_UNSIGNED_NUMERIC_TYPE(pCtx->inputType)) {
@ -842,7 +688,7 @@ static void avg_function(SqlFunctionCtx *pCtx) {
} }
} else { } else {
void *pData = GET_INPUT_DATA_LIST(pCtx); void *pData = GET_INPUT_DATA_LIST(pCtx);
if (pCtx->inputType == TSDB_DATA_TYPE_TINYINT) { if (pCtx->inputType == TSDB_DATA_TYPE_TINYINT) {
LIST_ADD_N(*pVal, pCtx, pData, int8_t, notNullElems, pCtx->inputType); LIST_ADD_N(*pVal, pCtx, pData, int8_t, notNullElems, pCtx->inputType);
} else if (pCtx->inputType == TSDB_DATA_TYPE_SMALLINT) { } else if (pCtx->inputType == TSDB_DATA_TYPE_SMALLINT) {
@ -865,18 +711,18 @@ static void avg_function(SqlFunctionCtx *pCtx) {
LIST_ADD_N(*pVal, pCtx, pData, uint64_t, notNullElems, pCtx->inputType); LIST_ADD_N(*pVal, pCtx, pData, uint64_t, notNullElems, pCtx->inputType);
} }
} }
if (!pCtx->hasNull) { if (!pCtx->hasNull) {
assert(notNullElems == pCtx->size); assert(notNullElems == pCtx->size);
} }
SET_VAL(pCtx, notNullElems, 1); SET_VAL(pCtx, notNullElems, 1);
pAvgInfo->num += notNullElems; pAvgInfo->num += notNullElems;
if (notNullElems > 0) { if (notNullElems > 0) {
//pResInfo->hasResult = DATA_SET_FLAG; //pResInfo->hasResult = DATA_SET_FLAG;
} }
// keep the data into the final output buffer for super table query since this execution may be the last one // keep the data into the final output buffer for super table query since this execution may be the last one
if (pCtx->stableQuery) { if (pCtx->stableQuery) {
memcpy(pCtx->pOutput, GET_ROWCELL_INTERBUF(pResInfo), sizeof(SAvgInfo)); memcpy(pCtx->pOutput, GET_ROWCELL_INTERBUF(pResInfo), sizeof(SAvgInfo));
@ -885,18 +731,18 @@ static void avg_function(SqlFunctionCtx *pCtx) {
static void avg_func_merge(SqlFunctionCtx *pCtx) { static void avg_func_merge(SqlFunctionCtx *pCtx) {
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx); SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
double *sum = (double*) pCtx->pOutput; double *sum = (double*) pCtx->pOutput;
char *input = GET_INPUT_DATA_LIST(pCtx); char *input = GET_INPUT_DATA_LIST(pCtx);
for (int32_t i = 0; i < pCtx->size; ++i, input += pCtx->inputBytes) { for (int32_t i = 0; i < pCtx->size; ++i, input += pCtx->inputBytes) {
SAvgInfo *pInput = (SAvgInfo *)input; SAvgInfo *pInput = (SAvgInfo *)input;
if (pInput->num == 0) { // current input is null if (pInput->num == 0) { // current input is null
continue; continue;
} }
SET_DOUBLE_VAL(sum, *sum + pInput->sum); SET_DOUBLE_VAL(sum, *sum + pInput->sum);
// keep the number of data into the temp buffer // keep the number of data into the temp buffer
*(int64_t *)GET_ROWCELL_INTERBUF(pResInfo) += pInput->num; *(int64_t *)GET_ROWCELL_INTERBUF(pResInfo) += pInput->num;
} }
@ -2735,18 +2581,6 @@ static void deriv_function(SqlFunctionCtx *pCtx) {
GET_RES_INFO(pCtx)->numOfRes += notNullElems; GET_RES_INFO(pCtx)->numOfRes += notNullElems;
} }
#define DIFF_IMPL(ctx, d, type) \
do { \
if ((ctx)->param[1].param.nType == INITIAL_VALUE_NOT_ASSIGNED) { \
(ctx)->param[1].param.nType = (ctx)->inputType; \
*(type *)&(ctx)->param[1].param.i = *(type *)(d); \
} else { \
*(type *)(ctx)->pOutput = *(type *)(d) - (*(type *)(&(ctx)->param[1].param.i)); \
*(type *)(&(ctx)->param[1].param.i) = *(type *)(d); \
*(int64_t *)(ctx)->pTsOutput = GET_TS_DATA(ctx, index); \
} \
} while (0);
// TODO difference in date column // TODO difference in date column
static void diff_function(SqlFunctionCtx *pCtx) { static void diff_function(SqlFunctionCtx *pCtx) {
void *data = GET_INPUT_DATA_LIST(pCtx); void *data = GET_INPUT_DATA_LIST(pCtx);
@ -4072,460 +3906,4 @@ int32_t functionCompatList[] = {
// tid_tag, derivative, blk_info // tid_tag, derivative, blk_info
6, 8, 7, 6, 8, 7,
}; };
#endif
SAggFunctionInfo aggFunc[35] = {{
// 0, count function does not invoke the finalize function
"count",
FUNCTION_TYPE_AGG,
FUNCTION_COUNT,
FUNCTION_COUNT,
BASIC_FUNC_SO,
function_setup,
count_function,
doFinalizer,
count_func_merge,
countRequired,
},
{
// 1
"sum",
FUNCTION_TYPE_AGG,
FUNCTION_SUM,
FUNCTION_SUM,
BASIC_FUNC_SO,
function_setup,
sum_function,
function_finalizer,
sum_func_merge,
statisRequired,
},
{
// 2
"avg",
FUNCTION_TYPE_AGG,
FUNCTION_AVG,
FUNCTION_AVG,
BASIC_FUNC_SO,
function_setup,
avg_function,
avg_finalizer,
avg_func_merge,
statisRequired,
},
{
// 3
"min",
FUNCTION_TYPE_AGG,
FUNCTION_MIN,
FUNCTION_MIN,
BASIC_FUNC_SO | FUNCSTATE_SELECTIVITY,
min_func_setup,
NULL,
function_finalizer,
min_func_merge,
statisRequired,
},
{
// 4
"max",
FUNCTION_TYPE_AGG,
FUNCTION_MAX,
FUNCTION_MAX,
BASIC_FUNC_SO | FUNCSTATE_SELECTIVITY,
max_func_setup,
NULL,
function_finalizer,
max_func_merge,
statisRequired,
},
{
// 5
"stddev",
FUNCTION_TYPE_AGG,
FUNCTION_STDDEV,
FUNCTION_STDDEV_DST,
FUNCSTATE_SO | FUNCSTATE_STREAM,
function_setup,
stddev_function,
stddev_finalizer,
noop1,
dataBlockRequired,
},
{
// 6
"percentile",
FUNCTION_TYPE_AGG,
FUNCTION_PERCT,
FUNCTION_INVALID_ID,
FUNCSTATE_SO | FUNCSTATE_STREAM,
percentile_function_setup,
percentile_function,
percentile_finalizer,
noop1,
dataBlockRequired,
},
{
// 7
"apercentile",
FUNCTION_TYPE_AGG,
FUNCTION_APERCT,
FUNCTION_APERCT,
FUNCSTATE_SO | FUNCSTATE_STREAM | FUNCSTATE_STABLE,
apercentile_function_setup,
apercentile_function,
apercentile_finalizer,
apercentile_func_merge,
dataBlockRequired,
},
{
// 8
"first",
FUNCTION_TYPE_AGG,
FUNCTION_FIRST,
FUNCTION_FIRST_DST,
BASIC_FUNC_SO | FUNCSTATE_SELECTIVITY,
function_setup,
first_function,
function_finalizer,
noop1,
firstFuncRequired,
},
{
// 9
"last",
FUNCTION_TYPE_AGG,
FUNCTION_LAST,
FUNCTION_LAST_DST,
BASIC_FUNC_SO | FUNCSTATE_SELECTIVITY,
function_setup,
last_function,
function_finalizer,
noop1,
lastFuncRequired,
},
{
// 10
"last_row",
FUNCTION_TYPE_AGG,
FUNCTION_LAST_ROW,
FUNCTION_LAST_ROW,
FUNCSTATE_SO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
first_last_function_setup,
last_row_function,
last_row_finalizer,
last_dist_func_merge,
dataBlockRequired,
},
{
// 11
"top",
FUNCTION_TYPE_AGG,
FUNCTION_TOP,
FUNCTION_TOP,
FUNCSTATE_MO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
top_bottom_function_setup,
top_function,
top_bottom_func_finalizer,
top_func_merge,
dataBlockRequired,
},
{
// 12
"bottom",
FUNCTION_TYPE_AGG,
FUNCTION_BOTTOM,
FUNCTION_BOTTOM,
FUNCSTATE_MO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
top_bottom_function_setup,
bottom_function,
top_bottom_func_finalizer,
bottom_func_merge,
dataBlockRequired,
},
{
// 13
"spread",
FUNCTION_TYPE_AGG,
FUNCTION_SPREAD,
FUNCTION_SPREAD,
BASIC_FUNC_SO,
spread_function_setup,
spread_function,
spread_function_finalizer,
spread_func_merge,
countRequired,
},
{
// 14
"twa",
FUNCTION_TYPE_AGG,
FUNCTION_TWA,
FUNCTION_TWA,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS,
twa_function_setup,
twa_function,
twa_function_finalizer,
twa_function_copy,
dataBlockRequired,
},
{
// 15
"leastsquares",
FUNCTION_TYPE_AGG,
FUNCTION_LEASTSQR,
FUNCTION_INVALID_ID,
FUNCSTATE_SO | FUNCSTATE_STREAM,
leastsquares_function_setup,
leastsquares_function,
leastsquares_finalizer,
noop1,
dataBlockRequired,
},
{
// 16
"dummy",
FUNCTION_TYPE_AGG,
FUNCTION_TS,
FUNCTION_TS,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS,
function_setup,
date_col_output_function,
doFinalizer,
copy_function,
noDataRequired,
},
{
// 17
"ts",
FUNCTION_TYPE_AGG,
FUNCTION_TS_DUMMY,
FUNCTION_TS_DUMMY,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS,
function_setup,
noop1,
doFinalizer,
copy_function,
dataBlockRequired,
},
{
// 18
"tag_dummy",
FUNCTION_TYPE_AGG,
FUNCTION_TAG_DUMMY,
FUNCTION_TAG_DUMMY,
BASIC_FUNC_SO,
function_setup,
tag_function,
doFinalizer,
copy_function,
noDataRequired,
},
{
// 19
"ts",
FUNCTION_TYPE_AGG,
FUNCTION_TS_COMP,
FUNCTION_TS_COMP,
FUNCSTATE_MO | FUNCSTATE_NEED_TS,
ts_comp_function_setup,
ts_comp_function,
ts_comp_finalize,
copy_function,
dataBlockRequired,
},
{
// 20
"tag",
FUNCTION_TYPE_AGG,
FUNCTION_TAG,
FUNCTION_TAG,
BASIC_FUNC_SO,
function_setup,
tag_function,
doFinalizer,
copy_function,
noDataRequired,
},
{//TODO this is a scala function
// 21, column project sql function
"colprj",
FUNCTION_TYPE_AGG,
FUNCTION_PRJ,
FUNCTION_PRJ,
BASIC_FUNC_MO | FUNCSTATE_NEED_TS,
function_setup,
col_project_function,
doFinalizer,
copy_function,
dataBlockRequired,
},
{
// 22, multi-output, tag function has only one result
"tagprj",
FUNCTION_TYPE_AGG,
FUNCTION_TAGPRJ,
FUNCTION_TAGPRJ,
BASIC_FUNC_MO,
function_setup,
tag_project_function,
doFinalizer,
copy_function,
noDataRequired,
},
{
// 23
"arithmetic",
FUNCTION_TYPE_AGG,
FUNCTION_ARITHM,
FUNCTION_ARITHM,
FUNCSTATE_MO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS,
function_setup,
arithmetic_function,
doFinalizer,
copy_function,
dataBlockRequired,
},
{
// 24
"diff",
FUNCTION_TYPE_AGG,
FUNCTION_DIFF,
FUNCTION_INVALID_ID,
FUNCSTATE_MO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
diff_function_setup,
diff_function,
doFinalizer,
noop1,
dataBlockRequired,
},
// distributed version used in two-stage aggregation processes
{
// 25
"first_dist",
FUNCTION_TYPE_AGG,
FUNCTION_FIRST_DST,
FUNCTION_FIRST_DST,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
first_last_function_setup,
first_dist_function,
function_finalizer,
first_dist_func_merge,
firstDistFuncRequired,
},
{
// 26
"last_dist",
FUNCTION_TYPE_AGG,
FUNCTION_LAST_DST,
FUNCTION_LAST_DST,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
first_last_function_setup,
last_dist_function,
function_finalizer,
last_dist_func_merge,
lastDistFuncRequired,
},
{
// 27
"stddev", // return table id and the corresponding tags for join match and subscribe
FUNCTION_TYPE_AGG,
FUNCTION_STDDEV_DST,
FUNCTION_AVG,
FUNCSTATE_SO | FUNCSTATE_STABLE,
function_setup,
NULL,
NULL,
NULL,
dataBlockRequired,
},
{
// 28
"interp",
FUNCTION_TYPE_AGG,
FUNCTION_INTERP,
FUNCTION_INTERP,
FUNCSTATE_SO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS ,
function_setup,
interp_function,
doFinalizer,
copy_function,
dataBlockRequired,
},
{
// 29
"rate",
FUNCTION_TYPE_AGG,
FUNCTION_RATE,
FUNCTION_RATE,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS,
rate_function_setup,
rate_function,
rate_finalizer,
rate_func_copy,
dataBlockRequired,
},
{
// 30
"irate",
FUNCTION_TYPE_AGG,
FUNCTION_IRATE,
FUNCTION_IRATE,
BASIC_FUNC_SO | FUNCSTATE_NEED_TS,
rate_function_setup,
irate_function,
rate_finalizer,
rate_func_copy,
dataBlockRequired,
},
{
// 31
"tbid", // return table id and the corresponding tags for join match and subscribe
FUNCTION_TYPE_AGG,
FUNCTION_TID_TAG,
FUNCTION_TID_TAG,
FUNCSTATE_MO | FUNCSTATE_STABLE,
function_setup,
noop1,
noop1,
noop1,
dataBlockRequired,
},
{ //32
"derivative", // return table id and the corresponding tags for join match and subscribe
FUNCTION_TYPE_AGG,
FUNCTION_DERIVATIVE,
FUNCTION_INVALID_ID,
FUNCSTATE_MO | FUNCSTATE_STABLE | FUNCSTATE_NEED_TS | FUNCSTATE_SELECTIVITY,
deriv_function_setup,
deriv_function,
doFinalizer,
noop1,
dataBlockRequired,
},
{
// 33
"block_dist", // return table id and the corresponding tags for join match and subscribe
FUNCTION_TYPE_AGG,
FUNCTION_BLKINFO,
FUNCTION_BLKINFO,
FUNCSTATE_SO | FUNCSTATE_STABLE,
function_setup,
blockInfo_func,
blockinfo_func_finalizer,
block_func_merge,
dataBlockRequired,
},
{
// 34
"cov", // return table id and the corresponding tags for join match and subscribe
FUNCTION_TYPE_AGG,
FUNCTION_COV,
FUNCTION_COV,
FUNCSTATE_SO | FUNCSTATE_STABLE,
function_setup,
sum_function,
function_finalizer,
sum_func_merge,
statisRequired,
}
};