Merge branch 'main' into feat/json-qb-tests

.github/workflows/integrationci.yaml

@@ -52,6 +52,7 @@ jobs:
           - ingestionkeys
           - rootuser
           - serviceaccount
+          - querier_json_body
         sqlstore-provider:
           - postgres
           - sqlite
@@ -61,7 +62,7 @@ jobs:
           - 25.5.6
           - 25.12.5
         schema-migrator-version:
-          - v0.142.0
+          - v0.144.2
         postgres-version:
           - 15
     if: |

pkg/instrumentation/sdk.go

@@ -14,7 +14,7 @@ import (
 	sdkmetric "go.opentelemetry.io/otel/metric"
 	sdkmetricnoop "go.opentelemetry.io/otel/metric/noop"
 	sdkresource "go.opentelemetry.io/otel/sdk/resource"
-	semconv "go.opentelemetry.io/otel/semconv/v1.40.0"
+	semconv "go.opentelemetry.io/otel/semconv/v1.39.0"
 	sdktrace "go.opentelemetry.io/otel/trace"
 )
 

pkg/querybuilder/collision.go

@@ -207,16 +207,23 @@ func AdjustKey(key *telemetrytypes.TelemetryFieldKey, keys map[string][]*telemet
 		indexes := []telemetrytypes.JSONDataTypeIndex{}
 		fieldContextsSeen := map[telemetrytypes.FieldContext]bool{}
 		dataTypesSeen := map[telemetrytypes.FieldDataType]bool{}
+		jsonTypesSeen := map[string]*telemetrytypes.JSONDataType{}
 		for _, matchingKey := range matchingKeys {
 			materialized = materialized && matchingKey.Materialized
 			fieldContextsSeen[matchingKey.FieldContext] = true
 			dataTypesSeen[matchingKey.FieldDataType] = true
+			if matchingKey.JSONDataType != nil {
+				jsonTypesSeen[matchingKey.JSONDataType.StringValue()] = matchingKey.JSONDataType
+			}
 			indexes = append(indexes, matchingKey.Indexes...)
 		}
 		for _, matchingKey := range contextPrefixedMatchingKeys {
 			materialized = materialized && matchingKey.Materialized
 			fieldContextsSeen[matchingKey.FieldContext] = true
 			dataTypesSeen[matchingKey.FieldDataType] = true
+			if matchingKey.JSONDataType != nil {
+				jsonTypesSeen[matchingKey.JSONDataType.StringValue()] = matchingKey.JSONDataType
+			}
 			indexes = append(indexes, matchingKey.Indexes...)
 		}
 		key.Materialized = materialized
@@ -241,6 +248,15 @@ func AdjustKey(key *telemetrytypes.TelemetryFieldKey, keys map[string][]*telemet
 				break
 			}
 		}
+
+		if len(jsonTypesSeen) == 1 && key.JSONDataType == nil {
+			// all matching keys have same JSON data type, use it
+			for _, jt := range jsonTypesSeen {
+				actions = append(actions, fmt.Sprintf("Adjusting key %s to have JSON data type %s", key, jt.StringValue()))
+				key.JSONDataType = jt
+				break
+			}
+		}
 	}
 
 	return actions

pkg/telemetrylogs/field_mapper.go

@@ -276,12 +276,11 @@ func (m *fieldMapper) FieldFor(ctx context.Context, tsStart, tsEnd uint64, key *
 					continue
 				}
 
-				if key.FieldDataType == telemetrytypes.FieldDataTypeUnspecified {
+				if key.JSONDataType == nil {
 					return "", qbtypes.ErrColumnNotFound
 				}
 
-				jdt := key.GetJSONDataType()
-				if key.KeyNameContainsArray() && !jdt.IsArray {
+				if key.KeyNameContainsArray() && !key.JSONDataType.IsArray {
 					return "", errors.NewInvalidInputf(errors.CodeInvalidInput, "FieldFor not supported for nested fields; only supported for flat paths (e.g. body.status.detail) and paths of Array type: %s(%s)", key.Name, key.FieldDataType)
 				}
 

pkg/telemetrylogs/json_stmt_builder_test.go

@@ -220,7 +220,7 @@ func TestJSONStmtBuilder_PrimitivePaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((LOWER(toString(dynamicElement(body_v2.`user.age`, 'Int64'))) LIKE LOWER(?)) AND has(JSONAllPaths(body_v2), 'user.age')) OR ((LOWER(dynamicElement(body_v2.`user.age`, 'String')) LIKE LOWER(?)) AND has(JSONAllPaths(body_v2), 'user.age')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "%25%", "%25%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `user.age` is ambiguous, found 2 different combinations of field context / data type: [name=user.age,context=body,datatype=int64 name=user.age,context=body,datatype=string]."},
+				Warnings:    []string{"Key `user.age` is ambiguous, found 2 different combinations of field context / data type: [name=user.age,context=body,datatype=int64,jsondatatype=Int64 name=user.age,context=body,datatype=string,jsondatatype=String]."},
 			},
 		},
 		{
@@ -414,7 +414,7 @@ func TestStatementBuilderListQueryBodyPromoted(t *testing.T) {
 			},
 			expected: TestExpected{
 				Args:     []any{uint64(1747945619), uint64(1747983448), "%1.65%", 1.65, "%1.65%", 1.65, "%1.65%", 1.65, "%1.65%", 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
+				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,materialized=true,jsondatatype=Array(Dynamic)]."},
 			},
 			expectedErr: nil,
 		},
@@ -441,7 +441,7 @@ func TestStatementBuilderListQueryBodyPromoted(t *testing.T) {
 			},
 			expected: TestExpected{
 				Args:     []any{uint64(1747945619), uint64(1747983448), "%true%", true, "%true%", true, "%true%", true, "%true%", true, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
+				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,materialized=true,jsondatatype=Array(Dynamic)]."},
 			},
 			expectedErr: nil,
 		},
@@ -455,7 +455,7 @@ func TestStatementBuilderListQueryBodyPromoted(t *testing.T) {
 			},
 			expected: TestExpected{
 				Args:     []any{uint64(1747945619), uint64(1747983448), "%passed%", "passed", "%passed%", "passed", "%passed%", "passed", "%passed%", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
+				Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,materialized=true,jsondatatype=Array(Dynamic)]."},
 			},
 			expectedErr: nil,
 		},
@@ -549,7 +549,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "((NOT arrayExists(`body_v2.education`-> toFloat64OrNull(dynamicElement(`body_v2.education`.`type`, 'String')) = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND (NOT arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`type`, 'Int64') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), int64(10001), int64(10001), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].type` is ambiguous, found 2 different combinations of field context / data type: [name=education[].type,context=body,datatype=string name=education[].type,context=body,datatype=int64]."},
+				Warnings:    []string{"Key `education[].type` is ambiguous, found 2 different combinations of field context / data type: [name=education[].type,context=body,datatype=string,jsondatatype=String name=education[].type,context=body,datatype=int64,jsondatatype=Int64]."},
 			},
 		},
 		{
@@ -576,7 +576,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "%1.65%", 1.65, "%1.65%", 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
+				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)]."},
 			},
 		},
 		{
@@ -585,7 +585,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "passed", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
+				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)]."},
 			},
 		},
 		{
@@ -594,7 +594,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toString(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> toString(x) = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "%passed%", "passed", "%passed%", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
+				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)]."},
 			},
 		},
 		{
@@ -603,7 +603,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) IN (?, ?), dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) IN (?, ?), arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), 1.65, 1.99, "1.65", "1.99", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
+				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)]."},
 			},
 		},
 		{
@@ -612,7 +612,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "((NOT arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND (NOT arrayExists(`body_v2.education`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "%1.65%", float64(1.65), "%1.65%", float64(1.65), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
+				Warnings:    []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)]."},
 			},
 		},
 		{
@@ -622,7 +622,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 				WhereClause: "(has(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?) OR has(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), 1.65, 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 				Warnings: []string{
-					"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic].",
+					"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,jsondatatype=Array(Nullable(Float64)) name=education[].parameters,context=body,datatype=[]dynamic,jsondatatype=Array(Dynamic)].",
 				},
 			},
 		},
@@ -702,7 +702,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			expected: TestExpected{
 				WhereClause: "(((arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(`body_v2.interests[].entities[].reviews`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`-> (arrayExists(x -> LOWER(toString(x)) LIKE LOWER(?), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(Int64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(Int64))'))), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata`.`positions`, 'Array(JSON(max_dynamic_types=0, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews[].entries`.`metadata`, 'Array(JSON(max_dynamic_types=1, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews`.`entries`, 'Array(JSON(max_dynamic_types=2, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities`.`reviews`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'interests')) OR ((arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(`body_v2.interests[].entities[].reviews`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`-> (arrayExists(x -> LOWER(x) LIKE LOWER(?), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(String))')) OR arrayExists(x -> toFloat64OrNull(x) = ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(String))'))), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata`.`positions`, 'Array(JSON(max_dynamic_types=0, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews[].entries`.`metadata`, 'Array(JSON(max_dynamic_types=1, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews`.`entries`, 'Array(JSON(max_dynamic_types=2, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities`.`reviews`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'interests')))",
 				Args:        []any{uint64(1747945619), uint64(1747983448), "%4%", float64(4), "%4%", float64(4), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
-				Warnings:    []string{"Key `interests[].entities[].reviews[].entries[].metadata[].positions[].ratings` is ambiguous, found 2 different combinations of field context / data type: [name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]int64 name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]string]."},
+				Warnings:    []string{"Key `interests[].entities[].reviews[].entries[].metadata[].positions[].ratings` is ambiguous, found 2 different combinations of field context / data type: [name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]int64,jsondatatype=Array(Nullable(Int64)) name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]string,jsondatatype=Array(Nullable(String))]."},
 			},
 		},
 		{
@@ -946,16 +946,16 @@ func buildTestTelemetryMetadataStore(t *testing.T, addIndexes bool) *telemetryty
 	mockMetadataStore := telemetrytypestest.NewMockMetadataStore()
 	mockMetadataStore.SetStaticFields(IntrinsicFields)
 	types, _ := telemetrytypes.TestJSONTypeSet()
-	for path, fieldDataTypes := range types {
-		for _, fdt := range fieldDataTypes {
+	for path, jsonTypes := range types {
+		for _, jsonType := range jsonTypes {
 			key := &telemetrytypes.TelemetryFieldKey{
 				Name:          path,
 				Signal:        telemetrytypes.SignalLogs,
 				FieldContext:  telemetrytypes.FieldContextBody,
-				FieldDataType: fdt,
+				FieldDataType: telemetrytypes.MappingJSONDataTypeToFieldDataType[jsonType],
+				JSONDataType:  &jsonType,
 			}
 			if addIndexes {
-				jsonType := telemetrytypes.MappingFieldDataTypeToJSONDataType[fdt]
 				idx := slices.IndexFunc(telemetrytypes.TestIndexedPaths, func(entry telemetrytypes.TestIndexedPathEntry) bool {
 					return entry.Path == path && entry.Type == jsonType
 				})

pkg/telemetrylogs/stmt_builder_test.go

@@ -875,6 +875,7 @@ func TestAdjustKey(t *testing.T) {
 			require.Equal(t, c.expectedKey.FieldContext, key.FieldContext, "field context should match")
 			require.Equal(t, c.expectedKey.FieldDataType, key.FieldDataType, "field data type should match")
 			require.Equal(t, c.expectedKey.Materialized, key.Materialized, "materialized should match")
+			require.Equal(t, c.expectedKey.JSONDataType, key.JSONDataType, "json data type should match")
 			require.Equal(t, c.expectedKey.Indexes, key.Indexes, "json exists should match")
 		})
 	}

pkg/telemetrymetadata/body_json_metadata.go

@@ -21,72 +21,133 @@ import (
 )
 
 var (
+	defaultPathLimit = 100 // Default limit to prevent full table scans
+
+	CodeUnknownJSONDataType     = errors.MustNewCode("unknown_json_data_type")
 	CodeFailLoadPromotedPaths   = errors.MustNewCode("fail_load_promoted_paths")
 	CodeFailCheckPathPromoted   = errors.MustNewCode("fail_check_path_promoted")
+	CodeFailIterateBodyJSONKeys = errors.MustNewCode("fail_iterate_body_json_keys")
+	CodeFailExtractBodyJSONKeys = errors.MustNewCode("fail_extract_body_json_keys")
 	CodeFailLoadLogsJSONIndexes = errors.MustNewCode("fail_load_logs_json_indexes")
 	CodeFailListJSONValues      = errors.MustNewCode("fail_list_json_values")
 	CodeFailScanJSONValue       = errors.MustNewCode("fail_scan_json_value")
 	CodeFailScanVariant         = errors.MustNewCode("fail_scan_variant")
+	CodeFailBuildJSONPathsQuery = errors.MustNewCode("fail_build_json_paths_query")
+	CodeNoPathsToQueryIndexes   = errors.MustNewCode("no_paths_to_query_indexes_provided")
 
 	CodeFailedToPrepareBatch = errors.MustNewCode("failed_to_prepare_batch_promoted_paths")
 	CodeFailedToSendBatch    = errors.MustNewCode("failed_to_send_batch_promoted_paths")
 	CodeFailedToAppendPath   = errors.MustNewCode("failed_to_append_path_promoted_paths")
 )
 
-// enrichBodyKeys enriches body-context keys with promoted path info, indexes,
-// and JSON access plans.
-// parentTypeCache contains parent array types (ArrayJSON/ArrayDynamic) pre-fetched in the main UNION query.
-func (t *telemetryMetaStore) enrichBodyKeys(ctx context.Context, keys []*telemetrytypes.TelemetryFieldKey, parentTypeCache map[string][]telemetrytypes.FieldDataType) error {
-	if len(keys) == 0 {
-		return nil
-	}
+// fetchBodyJSONPaths extracts body JSON paths from the path_types table
+// This function can be used by both JSONQueryBuilder and metadata extraction
+// uniquePathLimit: 0 for no limit, >0 for maximum number of unique paths to return
+//   - For startup load: set to 10000 to get top 10k unique paths
+//   - For lookup: set to 0 (no limit needed for single path)
+//   - For metadata API: set to desired pagination limit
+//
+// searchOperator: LIKE for pattern matching, EQUAL for exact match.
+func (t *telemetryMetaStore) fetchBodyJSONPaths(ctx context.Context,
+	fieldKeySelectors []*telemetrytypes.FieldKeySelector) ([]*telemetrytypes.TelemetryFieldKey, []string, bool, error) {
+	ctx = ctxtypes.NewContextWithCommentVals(ctx, map[string]string{
+		instrumentationtypes.TelemetrySignal:  telemetrytypes.SignalLogs.StringValue(),
+		instrumentationtypes.CodeNamespace:    "metadata",
+		instrumentationtypes.CodeFunctionName: "fetchBodyJSONPaths",
+	})
 
-	var filteredKeys []*telemetrytypes.TelemetryFieldKey
-	for _, key := range keys {
-		if key.FieldContext == telemetrytypes.FieldContextBody {
-			filteredKeys = append(filteredKeys, key)
+	query, args, limit := buildGetBodyJSONPathsQuery(fieldKeySelectors)
+	rows, err := t.telemetrystore.ClickhouseDB().Query(ctx, query, args...)
+	if err != nil {
+		return nil, nil, false, errors.WrapInternalf(err, CodeFailExtractBodyJSONKeys, "failed to extract body JSON keys")
+	}
+	defer rows.Close()
+
+	fieldKeys := []*telemetrytypes.TelemetryFieldKey{}
+	paths := []string{}
+	rowCount := 0
+	for rows.Next() {
+		var path string
+		var typesArray []string // ClickHouse returns array as []string
+		var lastSeen uint64
+
+		err = rows.Scan(&path, &typesArray, &lastSeen)
+		if err != nil {
+			return nil, nil, false, errors.WrapInternalf(err, CodeFailExtractBodyJSONKeys, "failed to scan body JSON key row")
 		}
+
+		for _, typ := range typesArray {
+			mapping, found := telemetrytypes.MappingStringToJSONDataType[typ]
+			if !found {
+				t.logger.ErrorContext(ctx, "failed to map type string to JSON data type", slog.String("type", typ), slog.String("path", path))
+				continue
+			}
+			fieldKeys = append(fieldKeys, &telemetrytypes.TelemetryFieldKey{
+				Name:          path,
+				Signal:        telemetrytypes.SignalLogs,
+				FieldContext:  telemetrytypes.FieldContextBody,
+				FieldDataType: telemetrytypes.MappingJSONDataTypeToFieldDataType[mapping],
+				JSONDataType:  &mapping,
+			})
+		}
+
+		paths = append(paths, path)
+		rowCount++
+	}
+	if rows.Err() != nil {
+		return nil, nil, false, errors.WrapInternalf(rows.Err(), CodeFailIterateBodyJSONKeys, "error iterating body JSON keys")
 	}
 
-	// collect paths for batch queries
-	paths := make([]string, 0, len(filteredKeys))
-	for _, key := range filteredKeys {
-		paths = append(paths, key.Name)
-	}
-
-	// fetch promoted paths
-	promoted, err := t.GetPromotedPaths(ctx, paths...)
-	if err != nil {
-		return err
-	}
-
-	// fetch JSON path indexes
-	indexes, err := t.getJSONPathIndexes(ctx, paths...)
-	if err != nil {
-		return err
-	}
-
-	// apply promoted/index metadata to keys
-	for _, key := range filteredKeys {
-		promotedKey := strings.Split(key.Name, telemetrytypes.ArraySep)[0]
-		key.Materialized = promoted[promotedKey]
-		key.Indexes = indexes[key.Name]
-	}
-
-	// build JSON access plans using the pre-fetched parent type cache
-	return t.buildJSONPlans(ctx, filteredKeys, parentTypeCache)
+	return fieldKeys, paths, rowCount <= limit, nil
 }
 
-// buildJSONPlans builds JSON access plans for the given keys
-// using the provided parent type cache (pre-fetched in the main UNION query).
-func (t *telemetryMetaStore) buildJSONPlans(_ context.Context, keys []*telemetrytypes.TelemetryFieldKey, typeCache map[string][]telemetrytypes.FieldDataType) error {
-	if len(keys) == 0 {
-		return nil
+func (t *telemetryMetaStore) buildBodyJSONPaths(ctx context.Context,
+	fieldKeySelectors []*telemetrytypes.FieldKeySelector) ([]*telemetrytypes.TelemetryFieldKey, bool, error) {
+
+	fieldKeys, paths, finished, err := t.fetchBodyJSONPaths(ctx, fieldKeySelectors)
+	if err != nil {
+		return nil, false, err
 	}
 
-	columnMeta := t.jsonColumnMetadata[telemetrytypes.SignalLogs][telemetrytypes.FieldContextBody]
+	promoted, err := t.GetPromotedPaths(ctx, paths...)
+	if err != nil {
+		return nil, false, err
+	}
+
+	indexes, err := t.getJSONPathIndexes(ctx, paths...)
+	if err != nil {
+		return nil, false, err
+	}
+
+	for _, fieldKey := range fieldKeys {
+		promotedKey := strings.Split(fieldKey.Name, telemetrytypes.ArraySep)[0]
+		fieldKey.Materialized = promoted[promotedKey]
+		fieldKey.Indexes = indexes[fieldKey.Name]
+	}
+
+	return fieldKeys, finished, t.buildJSONPlans(ctx, fieldKeys)
+}
+
+func (t *telemetryMetaStore) buildJSONPlans(ctx context.Context, keys []*telemetrytypes.TelemetryFieldKey) error {
+	parentSelectors := make([]*telemetrytypes.FieldKeySelector, 0, len(keys))
 	for _, key := range keys {
-		if err := key.SetJSONAccessPlan(columnMeta, typeCache); err != nil {
+		parentSelectors = append(parentSelectors, key.ArrayParentSelectors()...)
+	}
+
+	parentKeys, _, _, err := t.fetchBodyJSONPaths(ctx, parentSelectors)
+	if err != nil {
+		return err
+	}
+
+	typeCache := make(map[string][]telemetrytypes.JSONDataType)
+	for _, key := range parentKeys {
+		typeCache[key.Name] = append(typeCache[key.Name], *key.JSONDataType)
+	}
+
+	// build plans for keys now
+	for _, key := range keys {
+		err = key.SetJSONAccessPlan(t.jsonColumnMetadata[telemetrytypes.SignalLogs][telemetrytypes.FieldContextBody], typeCache)
+		if err != nil {
 			return err
 		}
 	}
@@ -94,6 +155,51 @@ func (t *telemetryMetaStore) buildJSONPlans(_ context.Context, keys []*telemetry
 	return nil
 }
 
+func buildGetBodyJSONPathsQuery(fieldKeySelectors []*telemetrytypes.FieldKeySelector) (string, []any, int) {
+	if len(fieldKeySelectors) == 0 {
+		return "", nil, defaultPathLimit
+	}
+	from := fmt.Sprintf("%s.%s", DBName, PathTypesTableName)
+
+	// Build a better query using GROUP BY to deduplicate at database level
+	// This aggregates all types per path and gets the max last_seen, then applies LIMIT
+	sb := sqlbuilder.Select(
+		"path",
+		"groupArray(DISTINCT type) AS types",
+		"max(last_seen) AS last_seen",
+	).From(from)
+
+	limit := 0
+	// Add search filter if provided
+	orClauses := []string{}
+	for _, fieldKeySelector := range fieldKeySelectors {
+		// replace [*] with []
+		fieldKeySelector.Name = strings.ReplaceAll(fieldKeySelector.Name, telemetrytypes.ArrayAnyIndex, telemetrytypes.ArraySep)
+		// Extract search text for body JSON keys
+		keyName := CleanPathPrefixes(fieldKeySelector.Name)
+		if fieldKeySelector.SelectorMatchType == telemetrytypes.FieldSelectorMatchTypeExact {
+			orClauses = append(orClauses, sb.Equal("path", keyName))
+		} else {
+			// Pattern matching for metadata API (defaults to LIKE behavior for other operators)
+			orClauses = append(orClauses, sb.ILike("path", fmt.Sprintf("%%%s%%", querybuilder.FormatValueForContains(keyName))))
+		}
+		limit += fieldKeySelector.Limit
+	}
+	sb.Where(sb.Or(orClauses...))
+	// Group by path to get unique paths with aggregated types
+	sb.GroupBy("path")
+
+	// Order by max last_seen to get most recent paths first
+	sb.OrderBy("last_seen DESC")
+	if limit == 0 {
+		limit = defaultPathLimit
+	}
+	sb.Limit(limit)
+
+	query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+	return query, args, limit
+}
+
 func (t *telemetryMetaStore) getJSONPathIndexes(ctx context.Context, paths ...string) (map[string][]telemetrytypes.JSONDataTypeIndex, error) {
 	filteredPaths := []string{}
 	for _, path := range paths {

pkg/telemetrymetadata/body_json_metadata_test.go

@@ -7,9 +7,99 @@ import (
 	"github.com/SigNoz/signoz-otel-collector/constants"
 	"github.com/SigNoz/signoz/pkg/querybuilder"
 	"github.com/SigNoz/signoz/pkg/telemetrylogs"
+	"github.com/SigNoz/signoz/pkg/types/telemetrytypes"
 	"github.com/stretchr/testify/require"
 )
 
+func TestBuildGetBodyJSONPathsQuery(t *testing.T) {
+	testCases := []struct {
+		name              string
+		fieldKeySelectors []*telemetrytypes.FieldKeySelector
+		expectedSQL       string
+		expectedArgs      []any
+		expectedLimit     int
+	}{
+
+		{
+			name: "Single search text with EQUAL operator",
+			fieldKeySelectors: []*telemetrytypes.FieldKeySelector{
+				{
+					Name:              "user.name",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeExact,
+				},
+			},
+			expectedSQL:   "SELECT path, groupArray(DISTINCT type) AS types, max(last_seen) AS last_seen FROM signoz_metadata.distributed_json_path_types WHERE (path = ?) GROUP BY path ORDER BY last_seen DESC LIMIT ?",
+			expectedArgs:  []any{"user.name", defaultPathLimit},
+			expectedLimit: defaultPathLimit,
+		},
+		{
+			name: "Single search text with LIKE operator",
+			fieldKeySelectors: []*telemetrytypes.FieldKeySelector{
+				{
+					Name:              "user",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeFuzzy,
+				},
+			},
+			expectedSQL:   "SELECT path, groupArray(DISTINCT type) AS types, max(last_seen) AS last_seen FROM signoz_metadata.distributed_json_path_types WHERE (LOWER(path) LIKE LOWER(?)) GROUP BY path ORDER BY last_seen DESC LIMIT ?",
+			expectedArgs:  []any{"%user%", 100},
+			expectedLimit: 100,
+		},
+		{
+			name: "Multiple search texts with EQUAL operator",
+			fieldKeySelectors: []*telemetrytypes.FieldKeySelector{
+				{
+					Name:              "user.name",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeExact,
+				},
+				{
+					Name:              "user.age",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeExact,
+				},
+			},
+			expectedSQL:   "SELECT path, groupArray(DISTINCT type) AS types, max(last_seen) AS last_seen FROM signoz_metadata.distributed_json_path_types WHERE (path = ? OR path = ?) GROUP BY path ORDER BY last_seen DESC LIMIT ?",
+			expectedArgs:  []any{"user.name", "user.age", defaultPathLimit},
+			expectedLimit: defaultPathLimit,
+		},
+		{
+			name: "Multiple search texts with LIKE operator",
+			fieldKeySelectors: []*telemetrytypes.FieldKeySelector{
+				{
+					Name:              "user",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeFuzzy,
+				},
+				{
+					Name:              "admin",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeFuzzy,
+				},
+			},
+			expectedSQL:   "SELECT path, groupArray(DISTINCT type) AS types, max(last_seen) AS last_seen FROM signoz_metadata.distributed_json_path_types WHERE (LOWER(path) LIKE LOWER(?) OR LOWER(path) LIKE LOWER(?)) GROUP BY path ORDER BY last_seen DESC LIMIT ?",
+			expectedArgs:  []any{"%user%", "%admin%", defaultPathLimit},
+			expectedLimit: defaultPathLimit,
+		},
+		{
+			name: "Search with Contains operator (should default to LIKE)",
+			fieldKeySelectors: []*telemetrytypes.FieldKeySelector{
+				{
+					Name:              "test",
+					SelectorMatchType: telemetrytypes.FieldSelectorMatchTypeFuzzy,
+				},
+			},
+			expectedSQL:   "SELECT path, groupArray(DISTINCT type) AS types, max(last_seen) AS last_seen FROM signoz_metadata.distributed_json_path_types WHERE (LOWER(path) LIKE LOWER(?)) GROUP BY path ORDER BY last_seen DESC LIMIT ?",
+			expectedArgs:  []any{"%test%", defaultPathLimit},
+			expectedLimit: defaultPathLimit,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			query, args, limit := buildGetBodyJSONPathsQuery(tc.fieldKeySelectors)
+			require.Equal(t, tc.expectedSQL, query)
+			require.Equal(t, tc.expectedArgs, args)
+			require.Equal(t, tc.expectedLimit, limit)
+		})
+	}
+}
+
 func TestBuildListLogsJSONIndexesQuery(t *testing.T) {
 	testCases := []struct {
 		name         string

pkg/telemetrymetadata/metadata.go

@@ -368,19 +368,6 @@ func (t *telemetryMetaStore) logsTblStatementToFieldKeys(ctx context.Context) ([
 	return materialisedKeys, nil
 }
 
-// logKeysUnionArm declares one arm of the UNION ALL in getLogsKeys.
-// All per-table variance is captured here so the loop body can stay uniform.
-type logKeysUnionArm struct {
-	shouldQuery        bool
-	fieldContext       telemetrytypes.FieldContext
-	table              string
-	dataTypeColumn     string                                    // column used in WHERE/GROUP BY
-	dataTypeSelectExpr string                                    // expression used in SELECT (may wrap with lower())
-	addBaseFilters     func(sb *sqlbuilder.SelectBuilder)        // mandatory WHERE filters (e.g., signal, field_context)
-	encodeDataType     func(telemetrytypes.FieldDataType) string // how to render a FieldDataType in WHERE values
-	extraOrBranch      func(sb *sqlbuilder.SelectBuilder) string // optional extra OR branch (e.g., body parent-types)
-}
-
 // getLogsKeys returns the keys from the spans that match the field selection criteria.
 func (t *telemetryMetaStore) getLogsKeys(ctx context.Context, fieldKeySelectors []*telemetrytypes.FieldKeySelector) ([]*telemetrytypes.TelemetryFieldKey, bool, error) {
 	ctx = ctxtypes.NewContextWithCommentVals(ctx, map[string]string{
@@ -410,149 +397,90 @@ func (t *telemetryMetaStore) getLogsKeys(ctx context.Context, fieldKeySelectors
 	// tables to query based on field selectors
 	queryAttributeTable := false
 	queryResourceTable := false
-	queryBodyTable := false
 
 	for _, selector := range fieldKeySelectors {
 		if selector.FieldContext == telemetrytypes.FieldContextUnspecified {
-			// unspecified context, query all tables
+			// unspecified context, query both tables
 			queryAttributeTable = true
 			queryResourceTable = true
-			queryBodyTable = true
 			break
 		} else if selector.FieldContext == telemetrytypes.FieldContextAttribute {
 			queryAttributeTable = true
 		} else if selector.FieldContext == telemetrytypes.FieldContextResource {
 			queryResourceTable = true
-		} else if selector.FieldContext == telemetrytypes.FieldContextBody && querybuilder.BodyJSONQueryEnabled {
-			queryBodyTable = true
 		}
 	}
 
-	// body keys are gated behind the feature flag
-	queryBodyTable = queryBodyTable && querybuilder.BodyJSONQueryEnabled
-
-	// pre-compute parent array path names from body selectors for JSON plan building;
-	// these will be fetched as a separate UNION arm filtered to ArrayJSON/ArrayDynamic only
-	parentPaths := make(map[string]bool)
-	if queryBodyTable {
-		for _, sel := range fieldKeySelectors {
-			if sel.FieldContext != telemetrytypes.FieldContextBody &&
-				sel.FieldContext != telemetrytypes.FieldContextUnspecified {
-				continue
-			}
-			if !strings.Contains(sel.Name, telemetrytypes.ArraySep) {
-				continue
-			}
-			key := &telemetrytypes.TelemetryFieldKey{
-				Name:         sel.Name,
-				Signal:       telemetrytypes.SignalLogs,
-				FieldContext: telemetrytypes.FieldContextBody,
-			}
-			for _, ps := range key.ArrayParentSelectors() {
-				parentPaths[ps.Name] = true
-			}
-		}
+	tablesToQuery := []struct {
+		fieldContext telemetrytypes.FieldContext
+		shouldQuery  bool
+	}{
+		{telemetrytypes.FieldContextAttribute, queryAttributeTable},
+		{telemetrytypes.FieldContextResource, queryResourceTable},
 	}
 
-	// Each UNION arm differs only in: table, data-type column name and SELECT
-	// expression (lower-wrapped for historical mixed-case in attr/resource),
-	// base WHERE filters, the per-selector data-type encoding, and (for body)
-	// an extra OR branch that fetches parent array types for JSON plan building.
-	// All other logic is shared by the loop below.
-	tablesToQuery := []logKeysUnionArm{
-		{
-			shouldQuery:        queryAttributeTable,
-			fieldContext:       telemetrytypes.FieldContextAttribute,
-			table:              t.logsDBName + "." + t.logAttributeKeysTblName,
-			dataTypeColumn:     "datatype",
-			dataTypeSelectExpr: "lower(datatype)",
-			addBaseFilters:     func(*sqlbuilder.SelectBuilder) {},
-			encodeDataType:     func(ft telemetrytypes.FieldDataType) string { return ft.TagDataType() },
-			extraOrBranch:      func(*sqlbuilder.SelectBuilder) string { return "" },
-		},
-		{
-			shouldQuery:        queryResourceTable,
-			fieldContext:       telemetrytypes.FieldContextResource,
-			table:              t.logsDBName + "." + t.logResourceKeysTblName,
-			dataTypeColumn:     "datatype",
-			dataTypeSelectExpr: "lower(datatype)",
-			addBaseFilters:     func(*sqlbuilder.SelectBuilder) {},
-			encodeDataType:     func(ft telemetrytypes.FieldDataType) string { return ft.TagDataType() },
-			extraOrBranch:      func(*sqlbuilder.SelectBuilder) string { return "" },
-		},
-		{
-			shouldQuery:        queryBodyTable,
-			fieldContext:       telemetrytypes.FieldContextBody,
-			table:              fmt.Sprintf("%s.%s", DBName, FieldKeysTable),
-			dataTypeColumn:     "field_data_type",
-			dataTypeSelectExpr: "field_data_type",
-			addBaseFilters: func(sb *sqlbuilder.SelectBuilder) {
-				sb.Where(sb.E("signal", telemetrytypes.SignalLogs.StringValue()))
-				sb.Where(sb.E("field_context", telemetrytypes.FieldContextBody.StringValue()))
-			},
-			encodeDataType: func(ft telemetrytypes.FieldDataType) string { return ft.StringValue() },
-			extraOrBranch: func(sb *sqlbuilder.SelectBuilder) string {
-				if len(parentPaths) == 0 {
-					return ""
-				}
-				names := make([]any, 0, len(parentPaths))
-				for n := range parentPaths {
-					names = append(names, n)
-				}
-				return sb.And(
-					sb.In("name", names...),
-					sb.In("field_data_type",
-						telemetrytypes.FieldDataTypeArrayDynamic.StringValue(),
-						telemetrytypes.FieldDataTypeArrayJSON.StringValue(),
-					),
-				)
-			},
-		},
-	}
-
-	for _, arm := range tablesToQuery {
-		if !arm.shouldQuery {
+	for _, table := range tablesToQuery {
+		if !table.shouldQuery {
 			continue
 		}
 
+		fieldContext := table.fieldContext
+
+		// table name based on field context
+		var tblName string
+		if fieldContext == telemetrytypes.FieldContextAttribute {
+			tblName = t.logsDBName + "." + t.logAttributeKeysTblName
+		} else {
+			tblName = t.logsDBName + "." + t.logResourceKeysTblName
+		}
+
 		sb := sqlbuilder.Select(
 			"name AS tag_key",
-			fmt.Sprintf("'%s' AS tag_type", arm.fieldContext.TagType()),
-			arm.dataTypeSelectExpr+" AS tag_data_type",
-			fmt.Sprintf("%d AS priority", getPriorityForContext(arm.fieldContext)),
-		).From(arm.table)
+			fmt.Sprintf("'%s' AS tag_type", fieldContext.TagType()),
+			"lower(datatype) AS tag_data_type", // in logs, we had some historical data with capital and small case
+			fmt.Sprintf(`%d AS priority`, getPriorityForContext(fieldContext)),
+		).From(tblName)
 
-		arm.addBaseFilters(sb)
+		var limit int
+		conds := []string{}
 
-		branches := []string{}
-		for _, sel := range fieldKeySelectors {
-			if sel.FieldContext != telemetrytypes.FieldContextUnspecified && sel.FieldContext != arm.fieldContext {
+		for _, fieldKeySelector := range fieldKeySelectors {
+			// Include this selector if:
+			// 1. It has unspecified context (matches all tables)
+			// 2. Its context matches the current table's context
+			if fieldKeySelector.FieldContext != telemetrytypes.FieldContextUnspecified &&
+				fieldKeySelector.FieldContext != fieldContext {
 				continue
 			}
-			parts := []string{}
-			if sel.SelectorMatchType == telemetrytypes.FieldSelectorMatchTypeExact {
-				parts = append(parts, sb.E("name", sel.Name))
+
+			// key part of the selector
+			fieldKeyConds := []string{}
+			if fieldKeySelector.SelectorMatchType == telemetrytypes.FieldSelectorMatchTypeExact {
+				fieldKeyConds = append(fieldKeyConds, sb.E("name", fieldKeySelector.Name))
 			} else {
-				parts = append(parts, sb.ILike("name", "%"+escapeForLike(sel.Name)+"%"))
+				fieldKeyConds = append(fieldKeyConds, sb.ILike("name", "%"+escapeForLike(fieldKeySelector.Name)+"%"))
 			}
-			if sel.FieldDataType != telemetrytypes.FieldDataTypeUnspecified {
-				parts = append(parts, sb.E(arm.dataTypeColumn, arm.encodeDataType(sel.FieldDataType)))
+
+			// now look at the field data type
+			if fieldKeySelector.FieldDataType != telemetrytypes.FieldDataTypeUnspecified {
+				fieldKeyConds = append(fieldKeyConds, sb.E("datatype", fieldKeySelector.FieldDataType.TagDataType()))
 			}
-			if len(parts) > 0 {
-				branches = append(branches, sb.And(parts...))
+
+			if len(fieldKeyConds) > 0 {
+				conds = append(conds, sb.And(fieldKeyConds...))
 			}
+			limit += fieldKeySelector.Limit
 		}
 
-		if extra := arm.extraOrBranch(sb); extra != "" {
-			branches = append(branches, extra)
+		if len(conds) > 0 {
+			sb.Where(sb.Or(conds...))
 		}
 
-		if len(branches) > 0 {
-			sb.Where(sb.Or(branches...))
+		sb.GroupBy("name", "datatype")
+		if limit == 0 {
+			limit = 1000
 		}
 
-		sb.GroupBy("name", arm.dataTypeColumn)
-
 		query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
 		queries = append(queries, query)
 		allArgs = append(allArgs, args...)
@@ -589,7 +517,6 @@ func (t *telemetryMetaStore) getLogsKeys(ctx context.Context, fieldKeySelectors
 	defer rows.Close()
 
 	keys := []*telemetrytypes.TelemetryFieldKey{}
-	parentTypeCache := make(map[string][]telemetrytypes.FieldDataType)
 	rowCount := 0
 	searchTexts := []string{}
 
@@ -613,17 +540,6 @@ func (t *telemetryMetaStore) getLogsKeys(ctx context.Context, fieldKeySelectors
 		if err != nil {
 			return nil, false, errors.Wrap(err, errors.TypeInternal, errors.CodeInternal, ErrFailedToGetLogsKeys.Error())
 		}
-
-		// body keys with ArrayJSON/ArrayDynamic types are internal container types
-		// used only for JSON access plan building; route to parentTypeCache, not to results
-		switch fieldDataType {
-		case telemetrytypes.FieldDataTypeArrayJSON, telemetrytypes.FieldDataTypeArrayDynamic:
-			if fieldContext == telemetrytypes.FieldContextBody && parentPaths[name] {
-				parentTypeCache[name] = append(parentTypeCache[name], fieldDataType)
-				continue
-			}
-		}
-
 		key, ok := mapOfKeys[name+";"+fieldContext.StringValue()+";"+fieldDataType.StringValue()]
 
 		// if there is no materialised column, create a key with the field context and data type
@@ -677,11 +593,13 @@ func (t *telemetryMetaStore) getLogsKeys(ctx context.Context, fieldKeySelectors
 		}
 	}
 
-	// enrich body keys with promoted paths, indexes, and JSON access plans
 	if querybuilder.BodyJSONQueryEnabled {
-		if err := t.enrichBodyKeys(ctx, keys, parentTypeCache); err != nil {
-			t.logger.ErrorContext(ctx, "failed to enrich body JSON keys", errors.Attr(err))
+		bodyJSONPaths, finished, err := t.buildBodyJSONPaths(ctx, fieldKeySelectors) // LIKE for pattern matching
+		if err != nil {
+			t.logger.ErrorContext(ctx, "failed to extract body JSON paths", errors.Attr(err))
 		}
+		keys = append(keys, bodyJSONPaths...)
+		complete = complete && finished
 	}
 
 	if _, err := t.updateColumnEvolutionMetadataForKeys(ctx, keys); err != nil {

pkg/telemetrymetadata/tables.go

@@ -1,11 +1,13 @@
 package telemetrymetadata
 
+import otelcollectorconst "github.com/SigNoz/signoz-otel-collector/constants"
+
 const (
 	DBName                           = "signoz_metadata"
 	AttributesMetadataTableName      = "distributed_attributes_metadata"
 	AttributesMetadataLocalTableName = "attributes_metadata"
 	ColumnEvolutionMetadataTableName = "distributed_column_evolution_metadata"
-	FieldKeysTable                   = "distributed_field_keys"
+	PathTypesTableName               = otelcollectorconst.DistributedPathTypesTable
 	// Column Evolution table stores promoted paths as (signal, column_name, field_context, field_name); see signoz-otel-collector metadata_migrations.
 	PromotedPathsTableName = "distributed_column_evolution_metadata"
 	SkipIndexTableName     = "system.data_skipping_indices"

pkg/types/telemetrytypes/field.go

@@ -37,6 +37,7 @@ type TelemetryFieldKey struct {
 	FieldContext  FieldContext  `json:"fieldContext,omitzero"`
 	FieldDataType FieldDataType `json:"fieldDataType,omitzero"`
 
+	JSONDataType *JSONDataType       `json:"-"`
 	JSONPlan     JSONAccessPlan      `json:"-"`
 	Indexes      []JSONDataTypeIndex `json:"-"`
 	Materialized bool                `json:"-"` // refers to promoted in case of body.... fields
@@ -79,12 +80,6 @@ func (f *TelemetryFieldKey) ArrayParentSelectors() []*FieldKeySelector {
 	return selectors
 }
 
-// GetJSONDataType derives the JSONDataType from FieldDataType.
-// Callers should check FieldDataType != FieldDataTypeUnspecified before calling.
-func (f *TelemetryFieldKey) GetJSONDataType() JSONDataType {
-	return MappingFieldDataTypeToJSONDataType[f.FieldDataType]
-}
-
 func (f TelemetryFieldKey) String() string {
 	var sb strings.Builder
 	fmt.Fprintf(&sb, "name=%s", f.Name)
@@ -97,6 +92,9 @@ func (f TelemetryFieldKey) String() string {
 	if f.Materialized {
 		sb.WriteString(",materialized=true")
 	}
+	if f.JSONDataType != nil {
+		fmt.Fprintf(&sb, ",jsondatatype=%s", f.JSONDataType.StringValue())
+	}
 	if len(f.Indexes) > 0 {
 		sb.WriteString(",indexes=[")
 		for i, index := range f.Indexes {
@@ -119,6 +117,7 @@ func (f TelemetryFieldKey) Text() string {
 func (f *TelemetryFieldKey) OverrideMetadataFrom(src *TelemetryFieldKey) {
 	f.FieldContext = src.FieldContext
 	f.FieldDataType = src.FieldDataType
+	f.JSONDataType = src.JSONDataType
 	f.Indexes = src.Indexes
 	f.Materialized = src.Materialized
 	f.JSONPlan = src.JSONPlan

pkg/types/telemetrytypes/field_datatype.go

@@ -31,7 +31,7 @@ var (
 	FieldDataTypeArrayInt64  = FieldDataType{valuer.NewString("[]int64")}
 	FieldDataTypeArrayNumber = FieldDataType{valuer.NewString("[]number")}
 
-	FieldDataTypeArrayJSON    = FieldDataType{valuer.NewString("[]object")}
+	FieldDataTypeArrayObject  = FieldDataType{valuer.NewString("[]object")}
 	FieldDataTypeArrayDynamic = FieldDataType{valuer.NewString("[]dynamic")}
 
 	// Map string representations to FieldDataType values
@@ -72,8 +72,6 @@ var (
 		"[]float64": FieldDataTypeArrayFloat64,
 		"[]number":  FieldDataTypeArrayNumber,
 		"[]bool":    FieldDataTypeArrayBool,
-		"[]json":    FieldDataTypeArrayJSON,
-		"[]dynamic": FieldDataTypeArrayDynamic,
 
 		// c-style array types
 		"string[]":  FieldDataTypeArrayString,
@@ -81,8 +79,6 @@ var (
 		"float64[]": FieldDataTypeArrayFloat64,
 		"number[]":  FieldDataTypeArrayNumber,
 		"bool[]":    FieldDataTypeArrayBool,
-		"json[]":    FieldDataTypeArrayJSON,
-		"dynamic[]": FieldDataTypeArrayDynamic,
 	}
 
 	fieldDataTypeToCHDataType = map[FieldDataType]string{

pkg/types/telemetrytypes/json_access_plan.go

@@ -43,7 +43,7 @@ type JSONAccessNode struct {
 	isRoot     bool // marked true for only body_v2 and body_promoted
 
 	// Precomputed type information (single source of truth)
-	AvailableTypes []FieldDataType
+	AvailableTypes []JSONDataType
 
 	// Array type branches (Array(JSON) vs Array(Dynamic))
 	Branches map[JSONAccessBranchType]*JSONAccessNode
@@ -106,7 +106,7 @@ type planBuilder struct {
 	paths      []string // cumulative paths for type cache lookups
 	segments   []string // individual path segments for node names
 	isPromoted bool
-	typeCache  map[string][]FieldDataType
+	typeCache  map[string][]JSONDataType
 }
 
 // buildPlan recursively builds the path plan tree.
@@ -155,14 +155,14 @@ func (pb *planBuilder) buildPlan(index int, parent *JSONAccessNode, isDynArrChil
 		MaxDynamicPaths: maxPaths,
 	}
 
-	hasJSON := slices.Contains(node.AvailableTypes, FieldDataTypeArrayJSON)
-	hasDynamic := slices.Contains(node.AvailableTypes, FieldDataTypeArrayDynamic)
+	hasJSON := slices.Contains(node.AvailableTypes, ArrayJSON)
+	hasDynamic := slices.Contains(node.AvailableTypes, ArrayDynamic)
 
 	// Configure terminal if this is the last part
 	if isTerminal {
 		node.TerminalConfig = &TerminalConfig{
 			Key:      pb.key,
-			ElemType: pb.key.GetJSONDataType(),
+			ElemType: *pb.key.JSONDataType,
 		}
 	} else {
 		var err error
@@ -185,7 +185,7 @@ func (pb *planBuilder) buildPlan(index int, parent *JSONAccessNode, isDynArrChil
 
 // buildJSONAccessPlan builds a tree structure representing the complete JSON path traversal
 // that precomputes all possible branches and their types.
-func (key *TelemetryFieldKey) SetJSONAccessPlan(columnInfo JSONColumnMetadata, typeCache map[string][]FieldDataType,
+func (key *TelemetryFieldKey) SetJSONAccessPlan(columnInfo JSONColumnMetadata, typeCache map[string][]JSONDataType,
 ) error {
 	// if path is empty, return nil
 	if key.Name == "" {

pkg/types/telemetrytypes/json_access_plan_test.go

@@ -19,11 +19,11 @@ const (
 // ============================================================================
 
 // makeKey creates a TelemetryFieldKey for testing.
-func makeKey(name string, dataType FieldDataType, materialized bool) *TelemetryFieldKey {
+func makeKey(name string, dataType JSONDataType, materialized bool) *TelemetryFieldKey {
 	return &TelemetryFieldKey{
-		Name:          name,
-		FieldDataType: dataType,
-		Materialized:  materialized,
+		Name:         name,
+		JSONDataType: &dataType,
+		Materialized: materialized,
 	}
 }
 
@@ -242,7 +242,7 @@ func TestPlanJSON_BasicStructure(t *testing.T) {
 	}{
 		{
 			name: "Simple path not promoted",
-			key:  makeKey("user.name", FieldDataTypeString, false),
+			key:  makeKey("user.name", String, false),
 			expectedYAML: fmt.Sprintf(`
 - name: user.name
   column: %s
@@ -255,7 +255,7 @@ func TestPlanJSON_BasicStructure(t *testing.T) {
 		},
 		{
 			name: "Simple path promoted",
-			key:  makeKey("user.name", FieldDataTypeString, true),
+			key:  makeKey("user.name", String, true),
 			expectedYAML: fmt.Sprintf(`
 - name: user.name
   column: %s
@@ -276,7 +276,7 @@ func TestPlanJSON_BasicStructure(t *testing.T) {
 		},
 		{
 			name:         "Empty path returns error",
-			key:          makeKey("", FieldDataTypeString, false),
+			key:          makeKey("", String, false),
 			expectErr:    true,
 			expectedYAML: "",
 		},
@@ -431,7 +431,7 @@ func TestPlanJSON_ArrayPaths(t *testing.T) {
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			key := makeKey(tt.path, FieldDataTypeString, false)
+			key := makeKey(tt.path, String, false)
 			err := key.SetJSONAccessPlan(JSONColumnMetadata{
 				BaseColumn:     bodyV2Column,
 				PromotedColumn: bodyPromotedColumn,
@@ -450,7 +450,7 @@ func TestPlanJSON_PromotedVsNonPromoted(t *testing.T) {
 	path := "education[].awards[].type"
 
 	t.Run("Non-promoted plan", func(t *testing.T) {
-		key := makeKey(path, FieldDataTypeString, false)
+		key := makeKey(path, String, false)
 		err := key.SetJSONAccessPlan(JSONColumnMetadata{
 			BaseColumn:     bodyV2Column,
 			PromotedColumn: bodyPromotedColumn,
@@ -493,7 +493,7 @@ func TestPlanJSON_PromotedVsNonPromoted(t *testing.T) {
 	})
 
 	t.Run("Promoted plan", func(t *testing.T) {
-		key := makeKey(path, FieldDataTypeString, true)
+		key := makeKey(path, String, true)
 		err := key.SetJSONAccessPlan(JSONColumnMetadata{
 			BaseColumn:     bodyV2Column,
 			PromotedColumn: bodyPromotedColumn,
@@ -666,12 +666,12 @@ func TestPlanJSON_EdgeCases(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			// Choose key type based on path; operator does not affect the tree shape asserted here.
-			keyType := FieldDataTypeString
+			keyType := String
 			switch tt.path {
 			case "education":
-				keyType = FieldDataTypeArrayJSON
+				keyType = ArrayJSON
 			case "education[].type":
-				keyType = FieldDataTypeString
+				keyType = String
 			}
 			key := makeKey(tt.path, keyType, false)
 			err := key.SetJSONAccessPlan(JSONColumnMetadata{
@@ -692,7 +692,7 @@ func TestPlanJSON_EdgeCases(t *testing.T) {
 func TestPlanJSON_TreeStructure(t *testing.T) {
 	types, _ := TestJSONTypeSet()
 	path := "education[].awards[].participated[].team[].branch"
-	key := makeKey(path, FieldDataTypeString, false)
+	key := makeKey(path, String, false)
 	err := key.SetJSONAccessPlan(JSONColumnMetadata{
 		BaseColumn:     bodyV2Column,
 		PromotedColumn: bodyPromotedColumn,

pkg/types/telemetrytypes/json_datatype.go

@@ -46,6 +46,14 @@ var MappingStringToJSONDataType = map[string]JSONDataType{
 	"Array(JSON)":              ArrayJSON,
 }
 
+var ScalerTypeToArrayType = map[JSONDataType]JSONDataType{
+	String:  ArrayString,
+	Int64:   ArrayInt64,
+	Float64: ArrayFloat64,
+	Bool:    ArrayBool,
+	Dynamic: ArrayDynamic,
+}
+
 var MappingFieldDataTypeToJSONDataType = map[FieldDataType]JSONDataType{
 	FieldDataTypeString:       String,
 	FieldDataTypeInt64:        Int64,
@@ -55,8 +63,18 @@ var MappingFieldDataTypeToJSONDataType = map[FieldDataType]JSONDataType{
 	FieldDataTypeArrayString:  ArrayString,
 	FieldDataTypeArrayInt64:   ArrayInt64,
 	FieldDataTypeArrayFloat64: ArrayFloat64,
-	FieldDataTypeArrayNumber:  ArrayFloat64,
 	FieldDataTypeArrayBool:    ArrayBool,
-	FieldDataTypeArrayDynamic: ArrayDynamic,
-	FieldDataTypeArrayJSON:    ArrayJSON,
+}
+
+var MappingJSONDataTypeToFieldDataType = map[JSONDataType]FieldDataType{
+	String:       FieldDataTypeString,
+	Int64:        FieldDataTypeInt64,
+	Float64:      FieldDataTypeFloat64,
+	Bool:         FieldDataTypeBool,
+	ArrayString:  FieldDataTypeArrayString,
+	ArrayInt64:   FieldDataTypeArrayInt64,
+	ArrayFloat64: FieldDataTypeArrayFloat64,
+	ArrayBool:    FieldDataTypeArrayBool,
+	ArrayDynamic: FieldDataTypeArrayDynamic,
+	ArrayJSON:    FieldDataTypeArrayObject,
 }

pkg/types/telemetrytypes/test_data.go

@@ -4,69 +4,69 @@ package telemetrytypes
 // Test JSON Type Set Data Setup
 // ============================================================================
 
-// TestJSONTypeSet returns a map of path->field data types for testing.
+// TestJSONTypeSet returns a map of path->types for testing.
 // This represents the type information available in the test JSON structure.
-func TestJSONTypeSet() (map[string][]FieldDataType, MetadataStore) {
-	types := map[string][]FieldDataType{
+func TestJSONTypeSet() (map[string][]JSONDataType, MetadataStore) {
+	types := map[string][]JSONDataType{
 
 		// ── user (primitives) ─────────────────────────────────────────────
-		"user.name":        {FieldDataTypeString},
-		"user.permissions": {FieldDataTypeArrayString},
-		"user.age":         {FieldDataTypeInt64, FieldDataTypeString}, // Int64/String ambiguity
-		"user.height":      {FieldDataTypeFloat64},
-		"user.active":      {FieldDataTypeBool}, // Bool — not IndexSupported
+		"user.name":        {String},
+		"user.permissions": {ArrayString},
+		"user.age":         {Int64, String}, // Int64/String ambiguity
+		"user.height":      {Float64},
+		"user.active":      {Bool}, // Bool — not IndexSupported
 
 		// Deeper non-array nesting (a.b.c — no array hops)
-		"user.address.zip": {FieldDataTypeInt64},
+		"user.address.zip": {Int64},
 
 		// ── education[] ───────────────────────────────────────────────────
 		// Pattern: x[].y
-		"education":              {FieldDataTypeArrayJSON},
-		"education[].name":       {FieldDataTypeString},
-		"education[].type":       {FieldDataTypeString, FieldDataTypeInt64},
-		"education[].year":       {FieldDataTypeInt64},
-		"education[].scores":     {FieldDataTypeArrayInt64},
-		"education[].parameters": {FieldDataTypeArrayFloat64, FieldDataTypeArrayDynamic},
+		"education":              {ArrayJSON},
+		"education[].name":       {String},
+		"education[].type":       {String, Int64},
+		"education[].year":       {Int64},
+		"education[].scores":     {ArrayInt64},
+		"education[].parameters": {ArrayFloat64, ArrayDynamic},
 
 		// Pattern: x[].y[]
-		"education[].awards": {FieldDataTypeArrayDynamic, FieldDataTypeArrayJSON},
+		"education[].awards": {ArrayDynamic, ArrayJSON},
 
 		// Pattern: x[].y[].z
-		"education[].awards[].name":     {FieldDataTypeString},
-		"education[].awards[].type":     {FieldDataTypeString},
-		"education[].awards[].semester": {FieldDataTypeInt64},
+		"education[].awards[].name":     {String},
+		"education[].awards[].type":     {String},
+		"education[].awards[].semester": {Int64},
 
 		// Pattern: x[].y[].z[]
-		"education[].awards[].participated": {FieldDataTypeArrayDynamic, FieldDataTypeArrayJSON},
+		"education[].awards[].participated": {ArrayDynamic, ArrayJSON},
 
 		// Pattern: x[].y[].z[].w
-		"education[].awards[].participated[].members": {FieldDataTypeArrayString},
+		"education[].awards[].participated[].members": {ArrayString},
 
 		// Pattern: x[].y[].z[].w[]
-		"education[].awards[].participated[].team": {FieldDataTypeArrayJSON},
+		"education[].awards[].participated[].team": {ArrayJSON},
 
 		// Pattern: x[].y[].z[].w[].v
-		"education[].awards[].participated[].team[].branch": {FieldDataTypeString},
+		"education[].awards[].participated[].team[].branch": {String},
 
 		// ── interests[] ───────────────────────────────────────────────────
-		"interests":                                                                 {FieldDataTypeArrayJSON},
-		"interests[].entities":                                                      {FieldDataTypeArrayJSON},
-		"interests[].entities[].reviews":                                            {FieldDataTypeArrayJSON},
-		"interests[].entities[].reviews[].entries":                                  {FieldDataTypeArrayJSON},
-		"interests[].entities[].reviews[].entries[].metadata":                       {FieldDataTypeArrayJSON},
-		"interests[].entities[].reviews[].entries[].metadata[].positions":           {FieldDataTypeArrayJSON},
-		"interests[].entities[].reviews[].entries[].metadata[].positions[].name":    {FieldDataTypeString},
-		"interests[].entities[].reviews[].entries[].metadata[].positions[].ratings": {FieldDataTypeArrayInt64, FieldDataTypeArrayString},
-		"http-events":                     {FieldDataTypeArrayJSON},
-		"http-events[].request-info.host": {FieldDataTypeString},
-		"ids":                             {FieldDataTypeArrayDynamic},
+		"interests":                                                                 {ArrayJSON},
+		"interests[].entities":                                                      {ArrayJSON},
+		"interests[].entities[].reviews":                                            {ArrayJSON},
+		"interests[].entities[].reviews[].entries":                                  {ArrayJSON},
+		"interests[].entities[].reviews[].entries[].metadata":                       {ArrayJSON},
+		"interests[].entities[].reviews[].entries[].metadata[].positions":           {ArrayJSON},
+		"interests[].entities[].reviews[].entries[].metadata[].positions[].name":    {String},
+		"interests[].entities[].reviews[].entries[].metadata[].positions[].ratings": {ArrayInt64, ArrayString},
+		"http-events":                     {ArrayJSON},
+		"http-events[].request-info.host": {String},
+		"ids":                             {ArrayDynamic},
 
 		// ── top-level primitives ──────────────────────────────────────────
-		"message":     {FieldDataTypeString},
-		"http-status": {FieldDataTypeInt64, FieldDataTypeString}, // hyphen in root key, ambiguous
+		"message":     {String},
+		"http-status": {Int64, String}, // hyphen in root key, ambiguous
 
 		// ── top-level nested objects (no array hops) ───────────────────────
-		"response.time-taken": {FieldDataTypeFloat64}, // hyphen inside nested key
+		"response.time-taken": {Float64}, // hyphen inside nested key
 	}
 
 	return types, nil

tests/integration/conftest.py

@@ -23,6 +23,7 @@ pytest_plugins = [
     "fixtures.notification_channel",
     "fixtures.alerts",
     "fixtures.cloudintegrations",
+    "fixtures.jsontypeexporter",
 ]
 
 

tests/integration/fixtures/jsontypeexporter.py

@@ -0,0 +1,437 @@
+"""
+Simpler version of jsontypeexporter for test fixtures.
+This exports JSON type metadata to the path_types table by parsing JSON bodies
+and extracting all paths with their types, similar to how the real jsontypeexporter works.
+"""
+
+import datetime
+import json
+from abc import ABC
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Callable,
+    Dict,
+    Generator,
+    List,
+    Optional,
+    Set,
+    Union,
+)
+
+import numpy as np
+import pytest
+
+from fixtures import types
+
+if TYPE_CHECKING:
+    pass
+
+
+class JSONPathType(ABC):
+    """Represents a JSON path with its type information"""
+
+    path: str
+    type: str
+    last_seen: np.uint64
+
+    def __init__(
+        self,
+        path: str,
+        type: str,  # pylint: disable=redefined-builtin
+        last_seen: Optional[datetime.datetime] = None,
+    ) -> None:
+        self.path = path
+        self.type = type
+        if last_seen is None:
+            last_seen = datetime.datetime.now()
+        self.last_seen = np.uint64(int(last_seen.timestamp() * 1e9))
+
+    def np_arr(self) -> np.array:
+        """Return path type data as numpy array for database insertion"""
+        return np.array([self.path, self.type, self.last_seen])
+
+
+# Constants matching jsontypeexporter
+ARRAY_SEPARATOR = "[]."  # Used in paths like "education[].name"
+ARRAY_SUFFIX = "[]"  # Used when traversing into array element objects
+
+
+def _infer_array_type_from_type_strings(types: List[str]) -> Optional[str]:
+    """
+    Infer array type from a list of pre-classified type strings.
+    Matches jsontypeexporter's inferArrayMask logic (v0.144.2+).
+
+    Type strings are: "JSON", "String", "Bool", "Float64", "Int64"
+
+    SuperTyping rules (matching Go inferArrayMask):
+    - JSON alone → Array(JSON)
+    - JSON + any primitive → Array(Dynamic)
+    - String alone → Array(Nullable(String)); String + other → Array(Dynamic)
+    - Float64 wins over Int64 and Bool
+    - Int64 wins over Bool
+    - Bool alone → Array(Nullable(Bool))
+    """
+    if len(types) == 0:
+        return None
+
+    unique = set(types)
+
+    has_json = "JSON" in unique
+    # hasPrimitive mirrors Go: (hasJSON && len(unique) > 1) || (!hasJSON && len(unique) > 0)
+    has_primitive = (has_json and len(unique) > 1) or (not has_json and len(unique) > 0)
+
+    if has_json:
+        if not has_primitive:
+            return "Array(JSON)"
+        return "Array(Dynamic)"
+
+    # ---- Primitive Type Resolution (Float > Int > Bool) ----
+    if "String" in unique:
+        if len(unique) > 1:
+            return "Array(Dynamic)"
+        return "Array(Nullable(String))"
+
+    if "Float64" in unique:
+        return "Array(Nullable(Float64))"
+    if "Int64" in unique:
+        return "Array(Nullable(Int64))"
+    if "Bool" in unique:
+        return "Array(Nullable(Bool))"
+
+    return "Array(Dynamic)"
+
+
+def _infer_array_type(elements: List[Any]) -> Optional[str]:
+    """
+    Infer array type from raw Python list elements.
+    Classifies each element then delegates to _infer_array_type_from_type_strings.
+    """
+    if len(elements) == 0:
+        return None
+
+    types = []
+    for elem in elements:
+        if elem is None:
+            continue
+        if isinstance(elem, dict):
+            types.append("JSON")
+        elif isinstance(elem, str):
+            types.append("String")
+        elif isinstance(elem, bool):  # must be before int (bool is subclass of int)
+            types.append("Bool")
+        elif isinstance(elem, float):
+            types.append("Float64")
+        elif isinstance(elem, int):
+            types.append("Int64")
+
+    return _infer_array_type_from_type_strings(types)
+
+
+def _python_type_to_clickhouse_type(value: Any) -> str:
+    """
+    Convert Python type to ClickHouse JSON type string.
+    Maps Python types to ClickHouse JSON data types.
+    """
+    if value is None:
+        return "String"  # Default for null values
+
+    if isinstance(value, bool):
+        return "Bool"
+    elif isinstance(value, int):
+        return "Int64"
+    elif isinstance(value, float):
+        return "Float64"
+    elif isinstance(value, str):
+        return "String"
+    elif isinstance(value, list):
+        # Use the sophisticated array type inference
+        array_type = _infer_array_type(value)
+        return array_type if array_type else "Array(Dynamic)"
+    elif isinstance(value, dict):
+        return "JSON"
+    else:
+        return "String"  # Default fallback
+
+
+def _extract_json_paths(
+    obj: Any,
+    current_path: str = "",
+    path_types: Optional[Dict[str, Set[str]]] = None,
+    level: int = 0,
+) -> Dict[str, Set[str]]:
+    """
+    Recursively extract all paths and their types from a JSON object.
+    Matches jsontypeexporter's analyzePValue logic.
+
+    Args:
+        obj: The JSON object to traverse
+        current_path: Current path being built (e.g., "user.name")
+        path_types: Dictionary mapping paths to sets of types found
+        level: Current nesting level (for depth limiting)
+
+    Returns:
+        Dictionary mapping paths to sets of type strings
+    """
+    if path_types is None:
+        path_types = {}
+
+    if obj is None:
+        if current_path:
+            if current_path not in path_types:
+                path_types[current_path] = set()
+            path_types[current_path].add("String")  # Null defaults to String
+        return path_types
+
+    if isinstance(obj, dict):
+        # For objects, add the object itself and recurse into keys
+        if current_path:
+            if current_path not in path_types:
+                path_types[current_path] = set()
+            path_types[current_path].add("JSON")
+
+        for key, value in obj.items():
+            # Build the path for this key
+            if current_path:
+                new_path = f"{current_path}.{key}"
+            else:
+                new_path = key
+
+            # Recurse into the value
+            _extract_json_paths(value, new_path, path_types, level + 1)
+
+    elif isinstance(obj, list):
+        # Skip empty arrays
+        if len(obj) == 0:
+            return path_types
+
+        # Collect types from array elements (matching Go: types := make([]pcommon.ValueType, 0, s.Len()))
+        types = []
+
+        for item in obj:
+            if isinstance(item, dict):
+                # When traversing into array element objects, use ArraySuffix ([])
+                # This matches: prefix+ArraySuffix in the Go code
+                # Example: if current_path is "education", we use "education[]" to traverse into objects
+                array_prefix = current_path + ARRAY_SUFFIX if current_path else ""
+                for key, value in item.items():
+                    if array_prefix:
+                        # Use array separator: education[].name
+                        array_path = f"{array_prefix}.{key}"
+                    else:
+                        array_path = key
+                    # Recurse without increasing level (matching Go behavior)
+                    _extract_json_paths(value, array_path, path_types, level)
+                types.append("JSON")
+            elif isinstance(item, list):
+                # Arrays inside arrays are not supported - skip the whole path
+                # Matching Go: e.logger.Error("arrays inside arrays are not supported!", ...); return nil
+                return path_types
+            elif isinstance(item, str):
+                types.append("String")
+            elif isinstance(item, bool):
+                types.append("Bool")
+            elif isinstance(item, float):
+                types.append("Float64")
+            elif isinstance(item, int):
+                types.append("Int64")
+
+        # Infer array type from collected types (matching Go: if mask := inferArrayMask(types); mask != 0)
+        if len(types) > 0:
+            array_type = _infer_array_type_from_type_strings(types)
+            if array_type and current_path:
+                if current_path not in path_types:
+                    path_types[current_path] = set()
+                path_types[current_path].add(array_type)
+
+    else:
+        # Primitive value (string, number, bool)
+        if current_path:
+            if current_path not in path_types:
+                path_types[current_path] = set()
+            obj_type = _python_type_to_clickhouse_type(obj)
+            path_types[current_path].add(obj_type)
+
+    return path_types
+
+
+def _parse_json_bodies_and_extract_paths(
+    json_bodies: List[str],
+    timestamp: Optional[datetime.datetime] = None,
+) -> List[JSONPathType]:
+    """
+    Parse JSON bodies and extract all paths with their types.
+    This mimics the behavior of jsontypeexporter.
+
+    Args:
+        json_bodies: List of JSON body strings to parse
+        timestamp: Timestamp to use for last_seen (defaults to now)
+
+    Returns:
+        List of JSONPathType objects with all discovered paths and types
+    """
+    if timestamp is None:
+        timestamp = datetime.datetime.now()
+
+    # Aggregate all paths and their types across all JSON bodies
+    all_path_types: Dict[str, Set[str]] = {}
+
+    for json_body in json_bodies:
+        try:
+            parsed = json.loads(json_body)
+            _extract_json_paths(parsed, "", all_path_types, level=0)
+        except (json.JSONDecodeError, TypeError):
+            # Skip invalid JSON
+            continue
+
+    # Convert to list of JSONPathType objects
+    # Each path can have multiple types, so we create one JSONPathType per type
+    path_type_objects: List[JSONPathType] = []
+    for path, types_set in all_path_types.items():
+        for type_str in types_set:
+            path_type_objects.append(
+                JSONPathType(path=path, type=type_str, last_seen=timestamp)
+            )
+
+    return path_type_objects
+
+
+@pytest.fixture(name="export_json_types", scope="function")
+def export_json_types(
+    clickhouse: types.TestContainerClickhouse,
+    request: pytest.FixtureRequest,  # To access migrator fixture
+) -> Generator[
+    Callable[[Union[List[JSONPathType], List[str], List[Any]]], None], Any, None
+]:
+    """
+    Fixture for exporting JSON type metadata to the path_types table.
+    This is a simpler version of jsontypeexporter for test fixtures.
+
+    The function can accept:
+    1. List of JSONPathType objects (manual specification)
+    2. List of JSON body strings (auto-extract paths)
+    3. List of Logs objects (extract from body_json field)
+
+    Usage examples:
+        # Manual specification
+        export_json_types([
+            JSONPathType(path="user.name", type="String"),
+            JSONPathType(path="user.age", type="Int64"),
+        ])
+
+        # Auto-extract from JSON strings
+        export_json_types([
+            '{"user": {"name": "alice", "age": 25}}',
+            '{"user": {"name": "bob", "age": 30}}',
+        ])
+
+        # Auto-extract from Logs objects
+        export_json_types(logs_list)
+    """
+    # Ensure migrator has run to create the table
+    try:
+        request.getfixturevalue("migrator")
+    except Exception:
+        # If migrator fixture is not available, that's okay - table might already exist
+        pass
+
+    def _export_json_types(
+        data: Union[
+            List[JSONPathType], List[str], List[Any]
+        ],  # List[Logs] but avoiding circular import
+    ) -> None:
+        """
+        Export JSON type metadata to signoz_metadata.distributed_json_path_types table.
+        This table stores path and type information for body JSON fields.
+        """
+        path_types: List[JSONPathType] = []
+
+        if len(data) == 0:
+            return
+
+        # Determine input type and convert to JSONPathType list
+        first_item = data[0]
+
+        if isinstance(first_item, JSONPathType):
+            # Already JSONPathType objects
+            path_types = data  # type: ignore
+        elif isinstance(first_item, str):
+            # List of JSON strings - parse and extract paths
+            path_types = _parse_json_bodies_and_extract_paths(data)  # type: ignore
+        else:
+            # Assume it's a list of Logs objects - extract body_v2
+            json_bodies: List[str] = []
+            for log in data:  # type: ignore
+                # Try to get body_v2 attribute
+                if hasattr(log, "body_v2") and log.body_v2:
+                    json_bodies.append(log.body_v2)
+                elif hasattr(log, "body") and log.body:
+                    # Fallback to body if body_v2 not available
+                    try:
+                        # Try to parse as JSON
+                        json.loads(log.body)
+                        json_bodies.append(log.body)
+                    except (json.JSONDecodeError, TypeError):
+                        pass
+
+            if json_bodies:
+                path_types = _parse_json_bodies_and_extract_paths(json_bodies)
+
+        if len(path_types) == 0:
+            return
+
+        clickhouse.conn.insert(
+            database="signoz_metadata",
+            table="distributed_json_path_types",
+            data=[path_type.np_arr() for path_type in path_types],
+            column_names=[
+                "path",
+                "type",
+                "last_seen",
+            ],
+        )
+
+    yield _export_json_types
+
+    # Cleanup - truncate the local table after tests (following pattern from logs fixture)
+    clickhouse.conn.query(
+        f"TRUNCATE TABLE signoz_metadata.json_path_types ON CLUSTER '{clickhouse.env['SIGNOZ_TELEMETRYSTORE_CLICKHOUSE_CLUSTER']}' SYNC"
+    )
+
+
+@pytest.fixture(name="export_promoted_paths", scope="function")
+def export_promoted_paths(
+    clickhouse: types.TestContainerClickhouse,
+    request: pytest.FixtureRequest,  # To access migrator fixture
+) -> Generator[Callable[[List[str]], None], Any, None]:
+    """
+    Fixture for exporting promoted JSON paths to the promoted paths table.
+    """
+    # Ensure migrator has run to create the table
+    try:
+        request.getfixturevalue("migrator")
+    except Exception:
+        # If migrator fixture is not available, that's okay - table might already exist
+        pass
+
+    def _export_promoted_paths(paths: List[str]) -> None:
+        if len(paths) == 0:
+            return
+
+        now_ms = int(datetime.datetime.now().timestamp() * 1000)
+        rows = [(path, now_ms) for path in paths]
+        clickhouse.conn.insert(
+            database="signoz_metadata",
+            table="distributed_json_promoted_paths",
+            data=rows,
+            column_names=[
+                "path",
+                "created_at",
+            ],
+        )
+
+    yield _export_promoted_paths
+
+    clickhouse.conn.query(
+        f"TRUNCATE TABLE signoz_metadata.json_promoted_paths ON CLUSTER '{clickhouse.env['SIGNOZ_TELEMETRYSTORE_CLICKHOUSE_CLUSTER']}' SYNC"
+    )

tests/integration/fixtures/logs.py

@@ -122,6 +122,8 @@ class Logs(ABC):
         resources: dict[str, Any] = {},
         attributes: dict[str, Any] = {},
         body: str = "default body",
+        body_v2: Optional[str] = None,
+        body_promoted: Optional[str] = None,
         severity_text: str = "INFO",
         trace_id: str = "",
         span_id: str = "",
@@ -167,6 +169,33 @@ class Logs(ABC):
         # Set body
         self.body = body
 
+        # Set body_v2 - if body is JSON, parse and stringify it, otherwise use empty string
+        # ClickHouse accepts String input for JSON column
+        if body_v2 is not None:
+            self.body_v2 = body_v2
+        else:
+            # Try to parse body as JSON; if successful use it directly,
+            # otherwise wrap as {"message": body} matching the normalize operator behavior.
+            try:
+                json.loads(body)
+                self.body_v2 = body
+            except (json.JSONDecodeError, TypeError):
+                self.body_v2 = json.dumps({"message": body})
+
+        # Set body_promoted - must be valid JSON
+        # Tests will explicitly pass promoted column's content, but we validate it
+        if body_promoted is not None:
+            # Validate that it's valid JSON
+            try:
+                json.loads(body_promoted)
+                self.body_promoted = body_promoted
+            except (json.JSONDecodeError, TypeError):
+                # If invalid, default to empty JSON object
+                self.body_promoted = "{}"
+        else:
+            # Default to empty JSON object (valid JSON)
+            self.body_promoted = "{}"
+
         # Process resources and attributes
         self.resources_string = {k: str(v) for k, v in resources.items()}
         self.resource_json = (
@@ -326,6 +355,8 @@ class Logs(ABC):
                 self.severity_text,
                 self.severity_number,
                 self.body,
+                self.body_v2,
+                self.body_promoted,
                 self.attributes_string,
                 self.attributes_number,
                 self.attributes_bool,
@@ -454,31 +485,53 @@ def insert_logs(
                 data=[resource_key.np_arr() for resource_key in resource_keys],
             )
 
+        # All columns in insertion order (must match Logs.np_arr() order)
+        all_column_names = [
+            "ts_bucket_start",
+            "resource_fingerprint",
+            "timestamp",
+            "observed_timestamp",
+            "id",
+            "trace_id",
+            "span_id",
+            "trace_flags",
+            "severity_text",
+            "severity_number",
+            "body",
+            "body_v2",
+            "body_promoted",
+            "attributes_string",
+            "attributes_number",
+            "attributes_bool",
+            "resources_string",
+            "scope_name",
+            "scope_version",
+            "scope_string",
+            "resource",
+        ]
+
+        # Check if body_v2 column exists (only present when ENABLE_LOGS_MIGRATIONS_V2 migration has run)
+        result = clickhouse.conn.query(
+            "SELECT count() FROM system.columns WHERE database = 'signoz_logs' AND table = 'logs_v2' AND name = 'body_v2'"
+        )
+        has_json_body = result.result_rows[0][0] > 0
+
+        if has_json_body:
+            column_names = all_column_names
+            data = [log.np_arr() for log in logs]
+        else:
+            json_body_cols = {"body_v2", "body_promoted"}
+            keep_indices = [
+                i for i, c in enumerate(all_column_names) if c not in json_body_cols
+            ]
+            column_names = [all_column_names[i] for i in keep_indices]
+            data = [log.np_arr()[keep_indices] for log in logs]
+
         clickhouse.conn.insert(
             database="signoz_logs",
             table="distributed_logs_v2",
-            data=[log.np_arr() for log in logs],
-            column_names=[
-                "ts_bucket_start",
-                "resource_fingerprint",
-                "timestamp",
-                "observed_timestamp",
-                "id",
-                "trace_id",
-                "span_id",
-                "trace_flags",
-                "severity_text",
-                "severity_number",
-                "body",
-                "attributes_string",
-                "attributes_number",
-                "attributes_bool",
-                "resources_string",
-                "scope_name",
-                "scope_version",
-                "scope_string",
-                "resource",
-            ],
+            data=data,
+            column_names=column_names,
         )
 
     yield _insert_logs

tests/integration/fixtures/migrator.py

@@ -1,3 +1,5 @@
+from typing import Optional
+
 import docker
 import pytest
 from testcontainers.core.container import Network
@@ -8,27 +10,32 @@ from fixtures.logger import setup_logger
 logger = setup_logger(__name__)
 
 
-@pytest.fixture(name="migrator", scope="package")
-def migrator(
+def create_migrator(
     network: Network,
     clickhouse: types.TestContainerClickhouse,
     request: pytest.FixtureRequest,
     pytestconfig: pytest.Config,
+    cache_key: str = "migrator",
+    env_overrides: Optional[dict] = None,
 ) -> types.Operation:
     """
-    Package-scoped fixture for running schema migrations.
+    Factory function for running schema migrations.
+    Accepts optional env_overrides to customize the migrator environment.
     """
 
     def create() -> None:
         version = request.config.getoption("--schema-migrator-version")
         client = docker.from_env()
 
+        environment = dict(env_overrides) if env_overrides else {}
+
         container = client.containers.run(
             image=f"signoz/signoz-schema-migrator:{version}",
             command=f"sync --replication=true --cluster-name=cluster --up= --dsn={clickhouse.env["SIGNOZ_TELEMETRYSTORE_CLICKHOUSE_DSN"]}",
             detach=True,
             auto_remove=False,
             network=network.id,
+            environment=environment,
         )
 
         result = container.wait()
@@ -47,6 +54,7 @@ def migrator(
             detach=True,
             auto_remove=False,
             network=network.id,
+            environment=environment,
         )
 
         result = container.wait()
@@ -59,7 +67,7 @@ def migrator(
 
         container.remove()
 
-        return types.Operation(name="migrator")
+        return types.Operation(name=cache_key)
 
     def delete(_: types.Operation) -> None:
         pass
@@ -70,9 +78,27 @@ def migrator(
     return dev.wrap(
         request,
         pytestconfig,
-        "migrator",
+        cache_key,
         lambda: types.Operation(name=""),
         create,
         delete,
         restore,
     )
+
+
+@pytest.fixture(name="migrator", scope="package")
+def migrator(
+    network: Network,
+    clickhouse: types.TestContainerClickhouse,
+    request: pytest.FixtureRequest,
+    pytestconfig: pytest.Config,
+) -> types.Operation:
+    """
+    Package-scoped fixture for running schema migrations.
+    """
+    return create_migrator(
+        network=network,
+        clickhouse=clickhouse,
+        request=request,
+        pytestconfig=pytestconfig,
+    )

tests/integration/src/querier_json_body/conftest.py

@@ -0,0 +1,70 @@
+import pytest
+from testcontainers.core.container import Network
+
+from fixtures import types
+from fixtures.migrator import create_migrator
+from fixtures.signoz import create_signoz
+
+UNSUPPORTED_CLICKHOUSE_VERSIONS = {"25.5.6"}
+
+
+def pytest_collection_modifyitems(
+    config: pytest.Config, items: list[pytest.Item]
+) -> None:
+    version = config.getoption("--clickhouse-version")
+    if version in UNSUPPORTED_CLICKHOUSE_VERSIONS:
+        skip = pytest.mark.skip(
+            reason=f"JSON body QB tests require ClickHouse > {version}"
+        )
+        for item in items:
+            item.add_marker(skip)
+
+
+@pytest.fixture(name="migrator", scope="package")
+def migrator_json(
+    network: Network,
+    clickhouse: types.TestContainerClickhouse,
+    request: pytest.FixtureRequest,
+    pytestconfig: pytest.Config,
+) -> types.Operation:
+    """
+    Package-scoped migrator with ENABLE_LOGS_MIGRATIONS_V2=1.
+    """
+    return create_migrator(
+        network=network,
+        clickhouse=clickhouse,
+        request=request,
+        pytestconfig=pytestconfig,
+        cache_key="migrator-json-body",
+        env_overrides={
+            "ENABLE_LOGS_MIGRATIONS_V2": "1",
+        },
+    )
+
+
+@pytest.fixture(name="signoz", scope="package")
+def signoz_json_body(
+    network: Network,
+    zeus: types.TestContainerDocker,
+    gateway: types.TestContainerDocker,
+    sqlstore: types.TestContainerSQL,
+    clickhouse: types.TestContainerClickhouse,
+    request: pytest.FixtureRequest,
+    pytestconfig: pytest.Config,
+) -> types.SigNoz:
+    """
+    Package-scoped fixture for SigNoz with BODY_JSON_QUERY_ENABLED=true.
+    """
+    return create_signoz(
+        network=network,
+        zeus=zeus,
+        gateway=gateway,
+        sqlstore=sqlstore,
+        clickhouse=clickhouse,
+        request=request,
+        pytestconfig=pytestconfig,
+        cache_key="signoz-json-body",
+        env_overrides={
+            "BODY_JSON_QUERY_ENABLED": "true",
+        },
+    )