2024-05-12 16:33:29 +02:00
|
|
|
package logstorage
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
"strings"
|
|
|
|
"unsafe"
|
|
|
|
)
|
|
|
|
|
|
|
|
type statsValues struct {
|
2024-05-22 21:01:20 +02:00
|
|
|
fields []string
|
|
|
|
limit uint64
|
2024-05-12 16:33:29 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
func (sv *statsValues) String() string {
|
2024-05-22 21:01:20 +02:00
|
|
|
s := "values(" + statsFuncFieldsToString(sv.fields) + ")"
|
2024-05-12 16:33:29 +02:00
|
|
|
if sv.limit > 0 {
|
|
|
|
s += fmt.Sprintf(" limit %d", sv.limit)
|
|
|
|
}
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
2024-05-20 04:08:30 +02:00
|
|
|
func (sv *statsValues) updateNeededFields(neededFields fieldsSet) {
|
2024-05-22 21:01:20 +02:00
|
|
|
updateNeededFieldsForStatsFunc(neededFields, sv.fields)
|
2024-05-12 16:33:29 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
func (sv *statsValues) newStatsProcessor() (statsProcessor, int) {
|
|
|
|
svp := &statsValuesProcessor{
|
|
|
|
sv: sv,
|
|
|
|
}
|
|
|
|
return svp, int(unsafe.Sizeof(*svp))
|
|
|
|
}
|
|
|
|
|
|
|
|
type statsValuesProcessor struct {
|
|
|
|
sv *statsValues
|
|
|
|
|
|
|
|
values []string
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) updateStatsForAllRows(br *blockResult) int {
|
|
|
|
if svp.limitReached() {
|
|
|
|
// Limit on the number of unique values has been reached
|
|
|
|
return 0
|
|
|
|
}
|
|
|
|
|
|
|
|
stateSizeIncrease := 0
|
2024-05-22 21:01:20 +02:00
|
|
|
fields := svp.sv.fields
|
|
|
|
if len(fields) == 0 {
|
2024-05-12 16:33:29 +02:00
|
|
|
for _, c := range br.getColumns() {
|
|
|
|
stateSizeIncrease += svp.updateStatsForAllRowsColumn(c, br)
|
|
|
|
}
|
|
|
|
} else {
|
2024-05-22 21:01:20 +02:00
|
|
|
for _, field := range fields {
|
2024-05-12 16:33:29 +02:00
|
|
|
c := br.getColumnByName(field)
|
|
|
|
stateSizeIncrease += svp.updateStatsForAllRowsColumn(c, br)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) updateStatsForAllRowsColumn(c *blockResultColumn, br *blockResult) int {
|
|
|
|
stateSizeIncrease := 0
|
|
|
|
if c.isConst {
|
2024-05-20 04:08:30 +02:00
|
|
|
v := strings.Clone(c.valuesEncoded[0])
|
2024-05-12 16:33:29 +02:00
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
|
|
|
|
values := svp.values
|
2024-09-25 16:16:53 +02:00
|
|
|
for i := 0; i < br.rowsLen; i++ {
|
2024-05-12 16:33:29 +02:00
|
|
|
values = append(values, v)
|
|
|
|
}
|
|
|
|
svp.values = values
|
|
|
|
|
2024-09-25 16:16:53 +02:00
|
|
|
stateSizeIncrease += br.rowsLen * int(unsafe.Sizeof(values[0]))
|
2024-05-12 16:33:29 +02:00
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
if c.valueType == valueTypeDict {
|
|
|
|
dictValues := make([]string, len(c.dictValues))
|
|
|
|
for i, v := range c.dictValues {
|
|
|
|
dictValues[i] = strings.Clone(v)
|
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
}
|
|
|
|
|
|
|
|
values := svp.values
|
2024-05-20 04:08:30 +02:00
|
|
|
for _, encodedValue := range c.getValuesEncoded(br) {
|
2024-05-12 16:33:29 +02:00
|
|
|
idx := encodedValue[0]
|
|
|
|
values = append(values, dictValues[idx])
|
|
|
|
}
|
|
|
|
svp.values = values
|
|
|
|
|
2024-09-25 16:16:53 +02:00
|
|
|
stateSizeIncrease += br.rowsLen * int(unsafe.Sizeof(values[0]))
|
2024-05-12 16:33:29 +02:00
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
values := svp.values
|
|
|
|
for _, v := range c.getValues(br) {
|
|
|
|
if len(values) == 0 || values[len(values)-1] != v {
|
|
|
|
v = strings.Clone(v)
|
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
}
|
|
|
|
values = append(values, v)
|
|
|
|
}
|
|
|
|
svp.values = values
|
|
|
|
|
2024-09-25 16:16:53 +02:00
|
|
|
stateSizeIncrease += br.rowsLen * int(unsafe.Sizeof(values[0]))
|
2024-05-12 16:33:29 +02:00
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) updateStatsForRow(br *blockResult, rowIdx int) int {
|
|
|
|
if svp.limitReached() {
|
|
|
|
// Limit on the number of unique values has been reached
|
|
|
|
return 0
|
|
|
|
}
|
|
|
|
|
|
|
|
stateSizeIncrease := 0
|
2024-05-22 21:01:20 +02:00
|
|
|
fields := svp.sv.fields
|
|
|
|
if len(fields) == 0 {
|
2024-05-12 16:33:29 +02:00
|
|
|
for _, c := range br.getColumns() {
|
|
|
|
stateSizeIncrease += svp.updateStatsForRowColumn(c, br, rowIdx)
|
|
|
|
}
|
|
|
|
} else {
|
2024-05-22 21:01:20 +02:00
|
|
|
for _, field := range fields {
|
2024-05-12 16:33:29 +02:00
|
|
|
c := br.getColumnByName(field)
|
|
|
|
stateSizeIncrease += svp.updateStatsForRowColumn(c, br, rowIdx)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) updateStatsForRowColumn(c *blockResultColumn, br *blockResult, rowIdx int) int {
|
|
|
|
stateSizeIncrease := 0
|
|
|
|
if c.isConst {
|
2024-05-20 04:08:30 +02:00
|
|
|
v := strings.Clone(c.valuesEncoded[0])
|
2024-05-12 16:33:29 +02:00
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
|
|
|
|
svp.values = append(svp.values, v)
|
|
|
|
stateSizeIncrease += int(unsafe.Sizeof(svp.values[0]))
|
|
|
|
|
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
if c.valueType == valueTypeDict {
|
|
|
|
// collect unique non-zero c.dictValues
|
2024-05-20 04:08:30 +02:00
|
|
|
valuesEncoded := c.getValuesEncoded(br)
|
|
|
|
dictIdx := valuesEncoded[rowIdx][0]
|
2024-05-12 16:33:29 +02:00
|
|
|
v := strings.Clone(c.dictValues[dictIdx])
|
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
|
|
|
|
svp.values = append(svp.values, v)
|
|
|
|
stateSizeIncrease += int(unsafe.Sizeof(svp.values[0]))
|
|
|
|
|
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
// collect unique values for the given rowIdx.
|
|
|
|
v := c.getValueAtRow(br, rowIdx)
|
|
|
|
v = strings.Clone(v)
|
|
|
|
stateSizeIncrease += len(v)
|
|
|
|
|
|
|
|
svp.values = append(svp.values, v)
|
|
|
|
stateSizeIncrease += int(unsafe.Sizeof(svp.values[0]))
|
|
|
|
|
|
|
|
return stateSizeIncrease
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) mergeState(sfp statsProcessor) {
|
|
|
|
if svp.limitReached() {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
src := sfp.(*statsValuesProcessor)
|
|
|
|
svp.values = append(svp.values, src.values...)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) finalizeStats() string {
|
|
|
|
items := svp.values
|
|
|
|
if len(items) == 0 {
|
|
|
|
return "[]"
|
|
|
|
}
|
|
|
|
|
|
|
|
if limit := svp.sv.limit; limit > 0 && uint64(len(items)) > limit {
|
|
|
|
items = items[:limit]
|
|
|
|
}
|
|
|
|
|
|
|
|
return marshalJSONArray(items)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (svp *statsValuesProcessor) limitReached() bool {
|
|
|
|
limit := svp.sv.limit
|
2024-09-29 10:50:25 +02:00
|
|
|
return limit > 0 && uint64(len(svp.values)) > limit
|
2024-05-12 16:33:29 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
func parseStatsValues(lex *lexer) (*statsValues, error) {
|
2024-05-22 21:01:20 +02:00
|
|
|
fields, err := parseStatsFuncFields(lex, "values")
|
2024-05-12 16:33:29 +02:00
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
|
|
|
}
|
|
|
|
sv := &statsValues{
|
2024-05-22 21:01:20 +02:00
|
|
|
fields: fields,
|
2024-05-12 16:33:29 +02:00
|
|
|
}
|
|
|
|
if lex.isKeyword("limit") {
|
|
|
|
lex.nextToken()
|
|
|
|
n, ok := tryParseUint64(lex.token)
|
|
|
|
if !ok {
|
|
|
|
return nil, fmt.Errorf("cannot parse 'limit %s' for 'values': %w", lex.token, err)
|
|
|
|
}
|
|
|
|
lex.nextToken()
|
|
|
|
sv.limit = n
|
|
|
|
}
|
|
|
|
return sv, nil
|
|
|
|
}
|