feat: remove unnessary extraction operation for unnest (#486)

Signed-off-by: ZhangJian He <shoothzj@gmail.com>

engine/immutable/logstore/unnest_func_test.go

@@ -24,49 +24,6 @@ import (
 	"github.com/openGemini/openGemini/lib/util/lifted/vm/protoparser/influx"
 )
 
-func TestUnnestMatchAll(t *testing.T) {
-	unnest := &influxql.Unnest{
-		Expr: &influxql.Call{
-			Name: "match_all",
-			Args: []influxql.Expr{&influxql.VarRef{Val: "location:([a-z]+)"}, &influxql.VarRef{Val: "content", Type: influxql.String}},
-		},
-		Aliases: []string{"key1"},
-		DstType: []influxql.DataType{influxql.String},
-	}
-	match, _ := NewUnnestMatchAll(unnest)
-	result := match.Get("location:test")
-	v, ok := result["key1"]
-	if !ok || v != "test" {
-		t.Fatal("get result error")
-	}
-	result = match.Get("type:test")
-	v, ok = result["key1"]
-	if !ok || v != "" {
-		t.Fatal("get nil result error")
-	}
-
-	matchNil, _ := NewUnnestMatchAll(&influxql.Unnest{
-		Expr:    &influxql.VarRef{},
-		Aliases: []string{"key1"},
-		DstType: []influxql.DataType{influxql.String},
-	})
-	if matchNil != nil {
-		t.Fatal("get nil match error")
-	}
-
-	_, err := NewUnnestMatchAll(&influxql.Unnest{
-		Expr: &influxql.Call{
-			Name: "match_all",
-			Args: []influxql.Expr{&influxql.VarRef{Val: "* | EXTRACT(content:\"type:(a-z:0-9]+\") AS(key1) | select count (key1) group by key1"}, &influxql.VarRef{Val: "content", Type: influxql.String}},
-		},
-		Aliases: []string{"key1"},
-		DstType: []influxql.DataType{influxql.String},
-	})
-	if err == nil {
-		t.Fatal("get err match")
-	}
-}
-
 func TestUnnestMatchAllOperator(t *testing.T) {
 	schema := record.Schemas{
 		record.Field{Type: influx.Field_Type_String, Name: "content"},

engine/immutable/unnest_func.go

@@ -16,50 +16,6 @@ limitations under the License.
 
 package immutable
 
-import (
-	"regexp"
-
-	"github.com/openGemini/openGemini/lib/util/lifted/influx/influxql"
-)
-
 type UnnestMatch interface {
 	Get([][]byte) [][]byte
 }
-
-type UnnestMatchAll struct {
-	unnest *influxql.Unnest
-	re     *regexp.Regexp
-}
-
-func NewUnnestMatchAll(unnest *influxql.Unnest) (*UnnestMatchAll, error) {
-	unnestExpr, ok := unnest.Expr.(*influxql.Call)
-	if !ok {
-		return nil, nil
-	}
-	var err error
-	reg, err := regexp.Compile(unnestExpr.Args[0].(*influxql.VarRef).Val)
-	if err != nil {
-		return nil, err
-	}
-	r := &UnnestMatchAll{
-		unnest: unnest,
-		re:     reg,
-	}
-
-	return r, nil
-}
-
-func (r *UnnestMatchAll) Get(s string) map[string]string {
-	result := make(map[string]string)
-	for _, v := range r.unnest.Aliases {
-		result[v] = ""
-	}
-	matchesAll := r.re.FindStringSubmatch(s)
-	if matchesAll == nil {
-		return result
-	}
-	for k, v := range r.unnest.Aliases {
-		result[v] = matchesAll[k+1]
-	}
-	return result
-}

lib/util/lifted/influx/httpd/handler_logstore.go

@@ -34,7 +34,6 @@ import (
 
 	"github.com/influxdata/influxdb/query"
 	"github.com/influxdata/influxdb/uuid"
-	logstore2 "github.com/openGemini/openGemini/engine/immutable/logstore"
 	"github.com/openGemini/openGemini/lib/bufferpool"
 	"github.com/openGemini/openGemini/lib/config"
 	"github.com/openGemini/openGemini/lib/cpu"
@@ -1996,26 +1995,9 @@ func (h *Handler) getQueryLogResult(resp *Response, logCond *influxql.Query, par
 	var count int64
 	var logs []map[string]interface{}
 	var unnest *influxql.Unnest
-	var unnestField string
 	if logCond != nil {
 		unnest = logCond.Statements[0].(*influxql.LogPipeStatement).Unnest
 	}
-	var unnestFunc *logstore2.UnnestMatchAll
-	if unnest != nil {
-		unnestExpr, ok := unnest.Expr.(*influxql.Call)
-		if !ok {
-			return 0, nil, fmt.Errorf("the type of unnest expr error")
-		}
-		unnestField = unnestExpr.Args[1].(*influxql.VarRef).Val
-		if unnestField == TAGS {
-			unnestField = Tag
-		}
-		var err error
-		unnestFunc, err = logstore2.NewUnnestMatchAll(unnest)
-		if err != nil {
-			return 0, nil, err
-		}
-	}
 	for i := range resp.Results {
 		for _, s := range resp.Results[i].Series {
 			for j := range s.Values {
@@ -2032,12 +2014,6 @@ func (h *Handler) getQueryLogResult(resp *Response, logCond *influxql.Query, par
 						}
 						continue
 					}
-					if unnest != nil && c == unnestField {
-						unnestResult := unnestFunc.Get(s.Values[j][id].(string))
-						for k, v := range unnestResult {
-							rec[k] = v
-						}
-					}
 					switch c {
 					case TIME:
 						v, _ := s.Values[j][id].(time.Time)
@@ -2049,8 +2025,12 @@ func (h *Handler) getQueryLogResult(resp *Response, logCond *influxql.Query, par
 					case CONTENT:
 						h.setRecord(rec, CONTENT, s.Values[j][id], para.Truncate)
 					default:
-						hasMore = true
-						h.setRecord(recMore, c, s.Values[j][id], para.Truncate)
+						if unnest != nil && unnest.IsUnnestField(c) {
+							h.setRecord(rec, c, s.Values[j][id], para.Truncate)
+						} else {
+							hasMore = true
+							h.setRecord(recMore, c, s.Values[j][id], para.Truncate)
+						}
 					}
 				}
 				rec[Cursor] = base64.StdEncoding.EncodeToString([]byte(strconv.FormatInt(currT, 10) + "|" + strconv.FormatInt(currSeqID, 10) + "^^"))

lib/util/lifted/influx/influxql/ast.go

@@ -7693,6 +7693,15 @@ func (u *Unnest) Clone() *Unnest {
 	return clone
 }
 
+func (u *Unnest) IsUnnestField(field string) bool {
+	for _, alias := range u.Aliases {
+		if alias == field {
+			return true
+		}
+	}
+	return false
+}
+
 func (u *Unnest) GetName() string {
 	return ""
 }