VictoriaMetrics/lib/logstorage/pipe_unpack_json.go

184 lines
4.3 KiB
Go
Raw Normal View History

2024-05-20 02:08:30 +00:00
package logstorage
import (
"fmt"
2024-05-22 19:01:20 +00:00
"slices"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
2024-05-20 02:08:30 +00:00
)
// pipeUnpackJSON processes '| unpack_json ...' pipe.
//
// See https://docs.victoriametrics.com/victorialogs/logsql/#unpack_json-pipe
type pipeUnpackJSON struct {
2024-05-22 19:01:20 +00:00
// fromField is the field to unpack json fields from
2024-05-20 02:08:30 +00:00
fromField string
2024-05-22 19:01:20 +00:00
// fields is an optional list of fields to extract from json.
//
// if it is empty, then all the fields are extracted.
fields []string
// resultPrefix is prefix to add to unpacked field names
2024-05-20 02:08:30 +00:00
resultPrefix string
2024-05-22 19:01:20 +00:00
// iff is an optional filter for skipping unpacking json
iff *ifFilter
2024-05-20 02:08:30 +00:00
}
func (pu *pipeUnpackJSON) String() string {
s := "unpack_json"
2024-05-22 19:01:20 +00:00
if pu.iff != nil {
s += " " + pu.iff.String()
}
2024-05-20 02:08:30 +00:00
if !isMsgFieldName(pu.fromField) {
s += " from " + quoteTokenIfNeeded(pu.fromField)
}
2024-05-22 19:01:20 +00:00
if len(pu.fields) > 0 {
s += " fields (" + fieldsToString(pu.fields) + ")"
}
2024-05-20 02:08:30 +00:00
if pu.resultPrefix != "" {
s += " result_prefix " + quoteTokenIfNeeded(pu.resultPrefix)
}
return s
}
func (pu *pipeUnpackJSON) updateNeededFields(neededFields, unneededFields fieldsSet) {
2024-05-22 19:01:20 +00:00
updateNeededFieldsForUnpackPipe(pu.fromField, pu.fields, pu.iff, neededFields, unneededFields)
}
func updateNeededFieldsForUnpackPipe(fromField string, outFields []string, iff *ifFilter, neededFields, unneededFields fieldsSet) {
2024-05-20 02:08:30 +00:00
if neededFields.contains("*") {
2024-05-22 19:01:20 +00:00
unneededFieldsOrig := unneededFields.clone()
unneededFieldsCount := 0
if len(outFields) > 0 {
for _, f := range outFields {
if unneededFieldsOrig.contains(f) {
unneededFieldsCount++
}
unneededFields.add(f)
}
}
if len(outFields) == 0 || unneededFieldsCount < len(outFields) {
unneededFields.remove(fromField)
if iff != nil {
unneededFields.removeFields(iff.neededFields)
}
}
2024-05-20 02:08:30 +00:00
} else {
2024-05-22 19:01:20 +00:00
neededFieldsOrig := neededFields.clone()
needFromField := len(outFields) == 0
if len(outFields) > 0 {
needFromField = false
for _, f := range outFields {
if neededFieldsOrig.contains(f) {
needFromField = true
}
neededFields.remove(f)
}
}
if needFromField {
neededFields.add(fromField)
if iff != nil {
neededFields.addFields(iff.neededFields)
}
}
2024-05-20 02:08:30 +00:00
}
}
func (pu *pipeUnpackJSON) newPipeProcessor(workersCount int, _ <-chan struct{}, _ func(), ppBase pipeProcessor) pipeProcessor {
2024-05-22 19:01:20 +00:00
unpackJSON := func(uctx *fieldsUnpackerContext, s string) {
if len(s) == 0 || s[0] != '{' {
// This isn't a JSON object
return
2024-05-20 02:08:30 +00:00
}
2024-05-22 19:01:20 +00:00
p := GetJSONParser()
err := p.ParseLogMessage(bytesutil.ToUnsafeBytes(s))
if err != nil {
for _, fieldName := range pu.fields {
uctx.addField(fieldName, "")
}
} else {
if len(pu.fields) == 0 {
for _, f := range p.Fields {
uctx.addField(f.Name, f.Value)
}
} else {
for _, fieldName := range pu.fields {
addedField := false
for _, f := range p.Fields {
if f.Name == fieldName {
uctx.addField(f.Name, f.Value)
addedField = true
break
}
}
if !addedField {
uctx.addField(fieldName, "")
}
}
2024-05-20 02:08:30 +00:00
}
}
2024-05-22 19:01:20 +00:00
PutJSONParser(p)
2024-05-20 02:08:30 +00:00
}
2024-05-22 19:01:20 +00:00
return newPipeUnpackProcessor(workersCount, unpackJSON, ppBase, pu.fromField, pu.resultPrefix, pu.iff)
2024-05-20 02:08:30 +00:00
}
func parsePipeUnpackJSON(lex *lexer) (*pipeUnpackJSON, error) {
if !lex.isKeyword("unpack_json") {
return nil, fmt.Errorf("unexpected token: %q; want %q", lex.token, "unpack_json")
}
lex.nextToken()
2024-05-22 19:01:20 +00:00
var iff *ifFilter
if lex.isKeyword("if") {
f, err := parseIfFilter(lex)
if err != nil {
return nil, err
}
iff = f
}
2024-05-20 02:08:30 +00:00
fromField := "_msg"
if lex.isKeyword("from") {
lex.nextToken()
f, err := parseFieldName(lex)
if err != nil {
return nil, fmt.Errorf("cannot parse 'from' field name: %w", err)
}
fromField = f
}
2024-05-22 19:01:20 +00:00
var fields []string
if lex.isKeyword("fields") {
lex.nextToken()
fs, err := parseFieldNamesInParens(lex)
if err != nil {
return nil, fmt.Errorf("cannot parse 'fields': %w", err)
}
fields = fs
if slices.Contains(fields, "*") {
fields = nil
}
}
2024-05-20 02:08:30 +00:00
resultPrefix := ""
if lex.isKeyword("result_prefix") {
lex.nextToken()
p, err := getCompoundToken(lex)
if err != nil {
return nil, fmt.Errorf("cannot parse 'result_prefix': %w", err)
}
resultPrefix = p
}
pu := &pipeUnpackJSON{
fromField: fromField,
2024-05-22 19:01:20 +00:00
fields: fields,
2024-05-20 02:08:30 +00:00
resultPrefix: resultPrefix,
2024-05-22 19:01:20 +00:00
iff: iff,
2024-05-20 02:08:30 +00:00
}
2024-05-22 19:01:20 +00:00
2024-05-20 02:08:30 +00:00
return pu, nil
}