VictoriaMetrics/lib/logstorage/pipe_format.go
Aliaksandr Valialkin 246c339e3d
lib/logstorage: read timestamps column when it is really needed during query execution
Previously timestamps column was read unconditionally on every query.
This could significantly slow down queries, which do not need reading this column
like in https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7070 .
2024-09-25 19:18:37 +02:00

286 lines
6.2 KiB
Go

package logstorage
import (
"fmt"
"math"
"unsafe"
"github.com/valyala/quicktemplate"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
)
// pipeFormat processes '| format ...' pipe.
//
// See https://docs.victoriametrics.com/victorialogs/logsql/#format-pipe
type pipeFormat struct {
formatStr string
steps []patternStep
resultField string
keepOriginalFields bool
skipEmptyResults bool
// iff is an optional filter for skipping the format func
iff *ifFilter
}
func (pf *pipeFormat) String() string {
s := "format"
if pf.iff != nil {
s += " " + pf.iff.String()
}
s += " " + quoteTokenIfNeeded(pf.formatStr)
if !isMsgFieldName(pf.resultField) {
s += " as " + quoteTokenIfNeeded(pf.resultField)
}
if pf.keepOriginalFields {
s += " keep_original_fields"
}
if pf.skipEmptyResults {
s += " skip_empty_results"
}
return s
}
func (pf *pipeFormat) canLiveTail() bool {
return true
}
func (pf *pipeFormat) updateNeededFields(neededFields, unneededFields fieldsSet) {
if neededFields.isEmpty() {
if pf.iff != nil {
neededFields.addFields(pf.iff.neededFields)
}
return
}
if neededFields.contains("*") {
if !unneededFields.contains(pf.resultField) {
if !pf.keepOriginalFields && !pf.skipEmptyResults {
unneededFields.add(pf.resultField)
}
if pf.iff != nil {
unneededFields.removeFields(pf.iff.neededFields)
}
for _, step := range pf.steps {
if step.field != "" {
unneededFields.remove(step.field)
}
}
}
} else {
if neededFields.contains(pf.resultField) {
if !pf.keepOriginalFields && !pf.skipEmptyResults {
neededFields.remove(pf.resultField)
}
if pf.iff != nil {
neededFields.addFields(pf.iff.neededFields)
}
for _, step := range pf.steps {
if step.field != "" {
neededFields.add(step.field)
}
}
}
}
}
func (pf *pipeFormat) optimize() {
pf.iff.optimizeFilterIn()
}
func (pf *pipeFormat) hasFilterInWithQuery() bool {
return pf.iff.hasFilterInWithQuery()
}
func (pf *pipeFormat) initFilterInValues(cache map[string][]string, getFieldValuesFunc getFieldValuesFunc) (pipe, error) {
iffNew, err := pf.iff.initFilterInValues(cache, getFieldValuesFunc)
if err != nil {
return nil, err
}
pfNew := *pf
pfNew.iff = iffNew
return &pfNew, nil
}
func (pf *pipeFormat) newPipeProcessor(workersCount int, _ <-chan struct{}, _ func(), ppNext pipeProcessor) pipeProcessor {
return &pipeFormatProcessor{
pf: pf,
ppNext: ppNext,
shards: make([]pipeFormatProcessorShard, workersCount),
}
}
type pipeFormatProcessor struct {
pf *pipeFormat
ppNext pipeProcessor
shards []pipeFormatProcessorShard
}
type pipeFormatProcessorShard struct {
pipeFormatProcessorShardNopad
// The padding prevents false sharing on widespread platforms with 128 mod (cache line size) = 0 .
_ [128 - unsafe.Sizeof(pipeFormatProcessorShardNopad{})%128]byte
}
type pipeFormatProcessorShardNopad struct {
bm bitmap
a arena
rc resultColumn
}
func (pfp *pipeFormatProcessor) writeBlock(workerID uint, br *blockResult) {
if br.rowsLen == 0 {
return
}
shard := &pfp.shards[workerID]
pf := pfp.pf
bm := &shard.bm
bm.init(br.rowsLen)
bm.setBits()
if iff := pf.iff; iff != nil {
iff.f.applyToBlockResult(br, bm)
if bm.isZero() {
pfp.ppNext.writeBlock(workerID, br)
return
}
}
shard.rc.name = pf.resultField
resultColumn := br.getColumnByName(pf.resultField)
for rowIdx := 0; rowIdx < br.rowsLen; rowIdx++ {
v := ""
if bm.isSetBit(rowIdx) {
v = shard.formatRow(pf, br, rowIdx)
if v == "" && pf.skipEmptyResults || pf.keepOriginalFields {
if vOrig := resultColumn.getValueAtRow(br, rowIdx); vOrig != "" {
v = vOrig
}
}
} else {
v = resultColumn.getValueAtRow(br, rowIdx)
}
shard.rc.addValue(v)
}
br.addResultColumn(&shard.rc)
pfp.ppNext.writeBlock(workerID, br)
shard.a.reset()
shard.rc.reset()
}
func (pfp *pipeFormatProcessor) flush() error {
return nil
}
func (shard *pipeFormatProcessorShard) formatRow(pf *pipeFormat, br *blockResult, rowIdx int) string {
b := shard.a.b
bLen := len(b)
for _, step := range pf.steps {
b = append(b, step.prefix...)
if step.field != "" {
c := br.getColumnByName(step.field)
v := c.getValueAtRow(br, rowIdx)
switch step.fieldOpt {
case "q":
b = quicktemplate.AppendJSONString(b, v, true)
case "time":
nsecs, ok := tryParseInt64(v)
if !ok {
b = append(b, v...)
continue
}
b = marshalTimestampRFC3339NanoString(b, nsecs)
case "duration":
nsecs, ok := tryParseInt64(v)
if !ok {
b = append(b, v...)
continue
}
b = marshalDurationString(b, nsecs)
case "ipv4":
ipNum, ok := tryParseUint64(v)
if !ok || ipNum > math.MaxUint32 {
b = append(b, v...)
continue
}
b = marshalIPv4String(b, uint32(ipNum))
default:
b = append(b, v...)
}
}
}
shard.a.b = b
return bytesutil.ToUnsafeString(b[bLen:])
}
func parsePipeFormat(lex *lexer) (*pipeFormat, error) {
if !lex.isKeyword("format") {
return nil, fmt.Errorf("unexpected token: %q; want %q", lex.token, "format")
}
lex.nextToken()
// parse optional if (...)
var iff *ifFilter
if lex.isKeyword("if") {
f, err := parseIfFilter(lex)
if err != nil {
return nil, err
}
iff = f
}
// parse format
formatStr, err := getCompoundToken(lex)
if err != nil {
return nil, fmt.Errorf("cannot read 'format': %w", err)
}
steps, err := parsePatternSteps(formatStr)
if err != nil {
return nil, fmt.Errorf("cannot parse 'pattern' %q: %w", formatStr, err)
}
// parse optional 'as ...` part
resultField := "_msg"
if lex.isKeyword("as") {
lex.nextToken()
field, err := parseFieldName(lex)
if err != nil {
return nil, fmt.Errorf("cannot parse result field after 'format %q as': %w", formatStr, err)
}
resultField = field
}
keepOriginalFields := false
skipEmptyResults := false
switch {
case lex.isKeyword("keep_original_fields"):
lex.nextToken()
keepOriginalFields = true
case lex.isKeyword("skip_empty_results"):
lex.nextToken()
skipEmptyResults = true
}
pf := &pipeFormat{
formatStr: formatStr,
steps: steps,
resultField: resultField,
keepOriginalFields: keepOriginalFields,
skipEmptyResults: skipEmptyResults,
iff: iff,
}
return pf, nil
}