fix: has function family revamp

pkg/querybuilder/where_clause_visitor.go

@@ -729,7 +729,11 @@ func (v *filterExpressionVisitor) VisitFunctionCall(ctx *grammar.FunctionCallCon
 		return ErrorConditionLiteral
 	}
 
-	value := params[1:]
+	value, err := normalizeFunctionValue(operator, functionName, params[1:])
+	if err != nil {
+		v.errors = append(v.errors, err.Error())
+		return ErrorConditionLiteral
+	}
 
 	conds, ok := v.buildConditions(key, matchingFieldKeys(key, v.fieldKeys), operator, value)
 	if !ok {
@@ -745,6 +749,44 @@ func (v *filterExpressionVisitor) VisitFunctionCall(ctx *grammar.FunctionCallCon
 	return v.builder.Or(conds...)
 }
 
+// normalizeFunctionValue validates and normalizes the value argument(s) of a has-family
+// function call, returning them in the wrapper slice the condition builder unwraps.
+//
+//   - has/hasToken take exactly one scalar value. More than one argument, or an array
+//     argument, is rejected rather than silently dropping the extras.
+//   - hasAny/hasAll take a set of values, supplied either as a single array literal
+//     (hasAny(k, ['a','b'])) or as several scalar arguments (hasAny(k, 'a', 'b')); the
+//     latter are folded into one list so no argument is silently ignored.
+func normalizeFunctionValue(operator qbtypes.FilterOperator, functionName string, valueParams []any) (any, error) {
+	switch operator {
+	case qbtypes.FilterOperatorHas, qbtypes.FilterOperatorHasToken:
+		if len(valueParams) != 1 {
+			return nil, fmt.Errorf("function `%s` expects exactly one value argument", functionName)
+		}
+		if _, isArray := valueParams[0].([]any); isArray {
+			return nil, fmt.Errorf("function `%s` expects a single scalar value, not an array", functionName)
+		}
+		return valueParams, nil
+	case qbtypes.FilterOperatorHasAny, qbtypes.FilterOperatorHasAll:
+		// A single array literal is already the value set.
+		if len(valueParams) == 1 {
+			if _, isArray := valueParams[0].([]any); isArray {
+				return valueParams, nil
+			}
+		}
+		// Otherwise fold the positional scalar arguments into one list.
+		values := make([]any, 0, len(valueParams))
+		for _, p := range valueParams {
+			if _, isArray := p.([]any); isArray {
+				return nil, fmt.Errorf("function `%s` expects either a single array literal or scalar values, not a mix of the two", functionName)
+			}
+			values = append(values, p)
+		}
+		return []any{values}, nil
+	}
+	return valueParams, nil
+}
+
 // VisitFunctionParamList handles the parameter list for function calls.
 func (v *filterExpressionVisitor) VisitFunctionParamList(ctx *grammar.FunctionParamListContext) any {
 	params := ctx.AllFunctionParam()

pkg/telemetrylogs/condition_builder.go

@@ -40,9 +40,8 @@ func isBodyJSONSearch(key *telemetrytypes.TelemetryFieldKey, columns []*schema.C
 	return false
 }
 
-// conditionForArrayFunction builds `has/hasAny/hasAll(<arrayFieldExpr>, value)` over a
-// body JSON array field. The field expression uses the JSON accessor (flag on) or
-// legacy string extraction (flag off); value[0] is the needle.
+// conditionForArrayFunction builds has/hasAny/hasAll over a body JSON path — via the JSON
+// access plan (flag on) or legacy typed extraction (flag off).
 func (c *conditionBuilder) conditionForArrayFunction(
 	ctx context.Context,
 	startNs, endNs uint64,
@@ -62,24 +61,48 @@ func (c *conditionBuilder) conditionForArrayFunction(
 		needle = args[0]
 	}
 
-	var fieldExpr string
 	if c.fl.BooleanOrEmpty(ctx, flagger.FeatureUseJSONBody, featuretypes.NewFlaggerEvaluationContext(valuer.UUID{})) {
-		fe, err := c.fm.FieldFor(ctx, startNs, endNs, key)
-		if err != nil {
-			return "", err
-		}
-		fieldExpr = fe
-	} else {
-		// legacy string-body path; value drives array-type inference (e.g. `[*]` paths)
-		fieldExpr, _ = GetBodyJSONKey(ctx, key, qbtypes.FilterOperatorUnknown, value)
+		// JSON access plan: data-type collision handling, nested array paths.
+		valueType, needle := InferDataType(needle, operator, key)
+		return NewJSONConditionBuilder(key, valueType).buildArrayFunctionCondition(operator, needle, sb)
 	}
 
-	return fmt.Sprintf("%s(%s, %s)", operator.FunctionName(), fieldExpr, sb.Var(needle)), nil
+	// legacy string-body path: type-matched array extraction, OR-ed with a scalar comparison
+	// for a scalar body value.
+	elemType := legacyElemType(needle)
+	arrayExpr := getBodyJSONArrayKey(key, elemType)
+	scalarExpr, scalarGuard, hasScalar := getBodyJSONScalarKey(key, elemType)
+	if list, ok := needle.([]any); ok {
+		vals := make([]any, len(list))
+		for i, v := range list {
+			vals[i] = legacyCoerceNeedle(v, elemType)
+		}
+		arrayCond := fmt.Sprintf("%s(%s, %s)", operator.FunctionName(), arrayExpr, sb.Var(vals))
+		if !hasScalar {
+			return arrayCond, nil
+		}
+		var membership string
+		if operator == qbtypes.FilterOperatorHasAll {
+			eqs := make([]string, len(vals))
+			for i, v := range vals {
+				eqs[i] = sb.E(scalarExpr, v)
+			}
+			membership = sb.And(eqs...)
+		} else {
+			membership = sb.In(scalarExpr, vals...)
+		}
+		return fmt.Sprintf("(%s OR %s)", arrayCond, sb.And(membership, scalarGuard)), nil
+	}
+	typedNeedle := legacyCoerceNeedle(needle, elemType)
+	arrayCond := fmt.Sprintf("%s(%s, %s)", operator.FunctionName(), arrayExpr, sb.Var(typedNeedle))
+	if !hasScalar {
+		return arrayCond, nil
+	}
+	return fmt.Sprintf("(%s OR %s)", arrayCond, sb.And(sb.E(scalarExpr, typedNeedle), scalarGuard)), nil
 }
 
-// conditionForHasToken builds `hasToken(LOWER(<bodyColumn>), LOWER(<needle>))`, a
-// full-text token search over the body column. It resolves the column from the key
-// name + use_json_body flag, validates the field/value, and tags errors with the doc URL.
+// conditionForHasToken builds a hasToken full-text search over the body column, resolving the
+// column from the key name + use_json_body flag.
 func (c *conditionBuilder) conditionForHasToken(
 	ctx context.Context,
 	key *telemetrytypes.TelemetryFieldKey,
@@ -92,28 +115,34 @@ func (c *conditionBuilder) conditionForHasToken(
 		needle = args[0]
 	}
 
-	// TODO(Tushar): thread orgID here to evaluate correctly
-	bodyJSONEnabled := c.fl.BooleanOrEmpty(ctx, flagger.FeatureUseJSONBody, featuretypes.NewFlaggerEvaluationContext(valuer.UUID{}))
-
-	columnName := LogsV2BodyColumn
-	if bodyJSONEnabled {
-		if key.Name != LogsV2BodyColumn && key.Name != bodyMessageField {
-			return "", errors.NewInvalidInputf(errors.CodeInvalidInput,
-				"function `hasToken` only supports body/body.message field as first parameter").WithUrl(hasTokenFunctionDocURL)
-		}
-		columnName = bodyMessageField
-	} else if key.Name != LogsV2BodyColumn {
-		return "", errors.NewInvalidInputf(errors.CodeInvalidInput,
-			"function `hasToken` only supports body field as first parameter").WithUrl(hasTokenFunctionDocURL)
-	}
-
 	// hasToken matches string tokens only.
 	if _, ok := needle.(string); !ok {
 		return "", errors.NewInvalidInputf(errors.CodeInvalidInput,
 			"function `hasToken` expects value parameter to be a string").WithUrl(hasTokenFunctionDocURL)
 	}
 
-	return fmt.Sprintf("hasToken(LOWER(%s), LOWER(%s))", columnName, sb.Var(needle)), nil
+	// TODO(Tushar): thread orgID here to evaluate correctly
+	bodyJSONEnabled := c.fl.BooleanOrEmpty(ctx, flagger.FeatureUseJSONBody, featuretypes.NewFlaggerEvaluationContext(valuer.UUID{}))
+
+	if !bodyJSONEnabled {
+		// legacy: token search over the plain body string column only.
+		if key.Name != LogsV2BodyColumn {
+			return "", errors.NewInvalidInputf(errors.CodeInvalidInput,
+				"function `hasToken` only supports body field as first parameter").WithUrl(hasTokenFunctionDocURL)
+		}
+		return fmt.Sprintf("hasToken(LOWER(%s), LOWER(%s))", LogsV2BodyColumn, sb.Var(needle)), nil
+	}
+
+	// JSON mode: a bare body/body.message key searches the body.message column; any other body
+	// field is a token search over its JSON string field, incl. strings nested in arrays.
+	if key.Name == LogsV2BodyColumn || key.Name == bodyMessageField {
+		return fmt.Sprintf("hasToken(LOWER(%s), LOWER(%s))", bodyMessageField, sb.Var(needle)), nil
+	}
+	if key.FieldContext == telemetrytypes.FieldContextBody {
+		return NewJSONConditionBuilder(key, telemetrytypes.FieldDataTypeString).buildTokenFunctionCondition(needle, sb)
+	}
+	return "", errors.NewInvalidInputf(errors.CodeInvalidInput,
+		"function `hasToken` only supports the body field or a body JSON string field as first parameter").WithUrl(hasTokenFunctionDocURL)
 }
 
 func (c *conditionBuilder) conditionFor(
@@ -124,8 +153,7 @@ func (c *conditionBuilder) conditionFor(
 	value any,
 	sb *sqlbuilder.SelectBuilder,
 ) (string, error) {
-	// hasToken is a token search over the body column resolved purely from the key
-	// name + flag, independent of column resolution, so handle it before anything else.
+	// hasToken resolves from the key name + flag alone (no column resolution), so handle it first.
 	if operator == qbtypes.FilterOperatorHasToken {
 		return c.conditionForHasToken(ctx, key, value, sb)
 	}
@@ -135,8 +163,7 @@ func (c *conditionBuilder) conditionFor(
 		return "", err
 	}
 
-	// has/hasAny/hasAll build `has(<arrayFieldExpr>, value)` over body JSON arrays
-	// rather than going through the normal operator paths, so handle them up front.
+	// has/hasAny/hasAll take the body-JSON path, not the normal operator paths.
 	if operator.IsArrayFunctionOperator() {
 		return c.conditionForArrayFunction(ctx, startNs, endNs, key, operator, value, columns, sb)
 	}
@@ -402,23 +429,6 @@ func (c *conditionBuilder) ConditionFor(
 		}
 	}
 
-	// has/hasAny/hasAll need an array field: in JSON-body mode drop non-array matches so a
-	// scalar errors clearly instead of failing at ClickHouse runtime (legacy mode skips this).
-	if operator.IsArrayFunctionOperator() &&
-		c.fl.BooleanOrEmpty(ctx, flagger.FeatureUseJSONBody, featuretypes.NewFlaggerEvaluationContext(valuer.UUID{})) {
-		arrayKeys := make([]*telemetrytypes.TelemetryFieldKey, 0, len(keys))
-		for _, k := range keys {
-			if k.FieldDataType.IsArray() {
-				arrayKeys = append(arrayKeys, k)
-			}
-		}
-		if len(arrayKeys) == 0 {
-			return nil, warnings, errors.NewInvalidInputf(errors.CodeInvalidInput,
-				"function `%s` expects key parameter to be an array field; no array fields found", operator.FunctionName())
-		}
-		keys = arrayKeys
-	}
-
 	conds := make([]string, 0, len(keys))
 	for _, k := range keys {
 		cond, err := c.conditionForKey(ctx, startNs, endNs, k, operator, value, sb)

pkg/telemetrylogs/filter_expr_logs_body_json_test.go

@@ -44,34 +44,66 @@ func TestFilterExprLogsBodyJSON(t *testing.T) {
 			category:              "json",
 			query:                 "has(body.requestor_list[*], 'index_service')",
 			shouldPass:            true,
-			expectedQuery:         `WHERE has(JSONExtract(JSON_QUERY(body, '$."requestor_list"[*]'), 'Array(String)'), ?)`,
-			expectedArgs:          []any{"index_service"},
+			expectedQuery:         `WHERE (has(JSONExtract(JSON_QUERY(body, '$."requestor_list"[*]'), 'Array(Nullable(String))'), ?) OR (JSON_VALUE(body, '$."requestor_list"') = ? AND JSONType(body, 'requestor_list') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{"index_service", "index_service"},
 			expectedErrorContains: "",
 		},
 		{
 			category:              "json",
 			query:                 "has(body.int_numbers[*], 2)",
 			shouldPass:            true,
-			expectedQuery:         `WHERE has(JSONExtract(JSON_QUERY(body, '$."int_numbers"[*]'), 'Array(Float64)'), ?)`,
-			expectedArgs:          []any{float64(2)},
+			expectedQuery:         `WHERE (has(JSONExtract(JSON_QUERY(body, '$."int_numbers"[*]'), 'Array(Nullable(Float64))'), ?) OR (JSONExtract(JSON_VALUE(body, '$."int_numbers"'), 'Nullable(Float64)') = ? AND JSONType(body, 'int_numbers') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{float64(2), float64(2)},
 			expectedErrorContains: "",
 		},
 		{
 			category:              "json",
 			query:                 "has(body.bool[*], true)",
 			shouldPass:            true,
-			expectedQuery:         `WHERE has(JSONExtract(JSON_QUERY(body, '$."bool"[*]'), 'Array(Bool)'), ?)`,
-			expectedArgs:          []any{true},
+			expectedQuery:         `WHERE (has(JSONExtract(JSON_QUERY(body, '$."bool"[*]'), 'Array(Nullable(String))'), ?) OR (JSON_VALUE(body, '$."bool"') = ? AND JSONType(body, 'bool') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{"true", "true"},
 			expectedErrorContains: "",
 		},
 		{
 			category:              "json",
 			query:                 "NOT has(body.nested_num[*].float_nums[*], 2.2)",
 			shouldPass:            true,
-			expectedQuery:         `WHERE NOT (has(JSONExtract(JSON_QUERY(body, '$."nested_num"[*]."float_nums"[*]'), 'Array(Float64)'), ?))`,
+			expectedQuery:         `WHERE NOT (has(JSONExtract(JSON_QUERY(body, '$."nested_num"[*]."float_nums"[*]'), 'Array(Nullable(Float64))'), ?))`,
 			expectedArgs:          []any{float64(2.2)},
 			expectedErrorContains: "",
 		},
+		{
+			category:              "json",
+			query:                 "has(body.tags, 'production')",
+			shouldPass:            true,
+			expectedQuery:         `WHERE (has(JSONExtract(JSON_QUERY(body, '$."tags"[*]'), 'Array(Nullable(String))'), ?) OR (JSON_VALUE(body, '$."tags"') = ? AND JSONType(body, 'tags') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{"production", "production"},
+			expectedErrorContains: "",
+		},
+		{
+			category:              "json",
+			query:                 "hasAny(body.tags, ['critical', 'test'])",
+			shouldPass:            true,
+			expectedQuery:         `WHERE (hasAny(JSONExtract(JSON_QUERY(body, '$."tags"[*]'), 'Array(Nullable(String))'), ?) OR (JSON_VALUE(body, '$."tags"') IN (?, ?) AND JSONType(body, 'tags') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{[]any{"critical", "test"}, "critical", "test"},
+			expectedErrorContains: "",
+		},
+		{
+			category:              "json",
+			query:                 "hasAll(body.tags, ['production', 'web'])",
+			shouldPass:            true,
+			expectedQuery:         `WHERE (hasAll(JSONExtract(JSON_QUERY(body, '$."tags"[*]'), 'Array(Nullable(String))'), ?) OR ((JSON_VALUE(body, '$."tags"') = ? AND JSON_VALUE(body, '$."tags"') = ?) AND JSONType(body, 'tags') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{[]any{"production", "web"}, "production", "web"},
+			expectedErrorContains: "",
+		},
+		{
+			category:              "json",
+			query:                 "has(body.ids, \"200\")",
+			shouldPass:            true,
+			expectedQuery:         `WHERE (has(JSONExtract(JSON_QUERY(body, '$."ids"[*]'), 'Array(Nullable(Int64))'), ?) OR (JSONExtract(JSON_VALUE(body, '$."ids"'), 'Nullable(Int64)') = ? AND JSONType(body, 'ids') NOT IN ('Array', 'Object', 'Null')))`,
+			expectedArgs:          []any{int64(200), int64(200)},
+			expectedErrorContains: "",
+		},
 		{
 			category:              "json",
 			query:                 "body.message = hello",

pkg/telemetrylogs/filter_expr_logs_test.go

@@ -1561,6 +1561,25 @@ func TestFilterExprLogs(t *testing.T) {
 			expectedArgs:          []any{"download"},
 			expectedErrorContains: "function `hasToken` expects value parameter to be a string",
 		},
+		// extra / mis-shaped value arguments are rejected, not silently dropped.
+		{
+			category:              "hasExtraArgs",
+			query:                 "has(body.tags[*], \"a\", \"b\")",
+			shouldPass:            false,
+			expectedErrorContains: "function `has` expects exactly one value argument",
+		},
+		{
+			category:              "hasArrayArg",
+			query:                 "has(body.tags[*], [\"a\", \"b\"])",
+			shouldPass:            false,
+			expectedErrorContains: "function `has` expects a single scalar value, not an array",
+		},
+		{
+			category:              "hasTokenExtraArgs",
+			query:                 "hasToken(body, \"a\", \"b\")",
+			shouldPass:            false,
+			expectedErrorContains: "function `hasToken` expects exactly one value argument",
+		},
 
 		// Basic materialized key
 		{

pkg/telemetrylogs/json_condition_builder.go

@@ -96,6 +96,18 @@ func applyNotCondition(operator qbtypes.FilterOperator) (bool, qbtypes.FilterOpe
 	return false, operator
 }
 
+// branchArrayExpr returns the ClickHouse array expression for a given array-type branch
+// at this hop. The JSON branch reads Array(JSON(...)) directly; the Dynamic branch filters
+// the Array(Dynamic) down to its JSON elements and maps them to JSON.
+func (c *jsonConditionBuilder) branchArrayExpr(node *telemetrytypes.JSONAccessNode, branch telemetrytypes.JSONAccessBranchType) string {
+	fieldPath := node.FieldPath()
+	if branch == telemetrytypes.BranchDynamic {
+		dynBaseExpr := fmt.Sprintf("dynamicElement(%s, 'Array(Dynamic)')", fieldPath)
+		return fmt.Sprintf("arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), %s))", dynBaseExpr)
+	}
+	return fmt.Sprintf("dynamicElement(%s, 'Array(JSON(max_dynamic_types=%d, max_dynamic_paths=%d))')", fieldPath, node.MaxDynamicTypes, node.MaxDynamicPaths)
+}
+
 // buildAccessNodeBranches builds conditions for each branch of the access node.
 func (c *jsonConditionBuilder) buildAccessNodeBranches(current *telemetrytypes.JSONAccessNode, operator qbtypes.FilterOperator, value any, sb *sqlbuilder.SelectBuilder) (string, error) {
 	if current == nil {
@@ -103,31 +115,15 @@ func (c *jsonConditionBuilder) buildAccessNodeBranches(current *telemetrytypes.J
 	}
 
 	currAlias := current.Alias()
-	fieldPath := current.FieldPath()
-	// Determine availability of Array(JSON) and Array(Dynamic) at this hop
-	hasArrayJSON := current.Branches[telemetrytypes.BranchJSON] != nil
-	hasArrayDynamic := current.Branches[telemetrytypes.BranchDynamic] != nil
-
-	// Then, at this hop, compute child per branch and wrap
+	// At this hop, compute the child condition per array branch (JSON before Dynamic) and
+	// wrap each in arrayExists over the corresponding array expression.
 	branches := make([]string, 0, 2)
-	if hasArrayJSON {
-		jsonArrayExpr := fmt.Sprintf("dynamicElement(%s, 'Array(JSON(max_dynamic_types=%d, max_dynamic_paths=%d))')", fieldPath, current.MaxDynamicTypes, current.MaxDynamicPaths)
-		childGroupJSON, err := c.recurseArrayHops(current.Branches[telemetrytypes.BranchJSON], operator, value, sb)
+	for _, branch := range current.BranchesInOrder() {
+		childGroup, err := c.recurseArrayHops(current.Branches[branch], operator, value, sb)
 		if err != nil {
 			return "", err
 		}
-		branches = append(branches, fmt.Sprintf("arrayExists(%s-> %s, %s)", currAlias, childGroupJSON, jsonArrayExpr))
-	}
-	if hasArrayDynamic {
-		dynBaseExpr := fmt.Sprintf("dynamicElement(%s, 'Array(Dynamic)')", fieldPath)
-		dynFilteredExpr := fmt.Sprintf("arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), %s))", dynBaseExpr)
-
-		// Create the Query for Dynamic array
-		childGroupDyn, err := c.recurseArrayHops(current.Branches[telemetrytypes.BranchDynamic], operator, value, sb)
-		if err != nil {
-			return "", err
-		}
-		branches = append(branches, fmt.Sprintf("arrayExists(%s-> %s, %s)", currAlias, childGroupDyn, dynFilteredExpr))
+		branches = append(branches, fmt.Sprintf("arrayExists(%s-> %s, %s)", currAlias, childGroup, c.branchArrayExpr(current, branch)))
 	}
 
 	if len(branches) == 1 {
@@ -309,6 +305,169 @@ func (c *jsonConditionBuilder) buildArrayMembershipCondition(node *telemetrytype
 	return fmt.Sprintf("arrayExists(%s -> %s, %s)", key, op, arrayExpr), nil
 }
 
+// buildArrayFunctionCondition builds a has/hasAny/hasAll condition over a body JSON path,
+// with contains-all semantics uniform across every leaf shape:
+//   - has(v)         = the path HAS v
+//   - hasAny([v...]) = the path has ANY listed value  (OR of has)
+//   - hasAll([v...]) = the path has ALL listed values (AND of has)
+//
+// "the path has v" is an existential match resolved per leaf shape: for an array-typed leaf
+// (top-level `body.tags`, or nested `body.education[].scores`) it is native membership; for a
+// scalar leaf — whether reached through an array hop (`body.items[].sku`) or a plain scalar
+// path (`body.level`) — it is `<elem> = v`, wrapped in arrayExists over any array hops. So
+// `hasAll(body.education[].name, ['a','b'])` = "some element is a AND some element is b", and
+// for a plain scalar hasAll collapses to has (a one-element set can hold at most one value).
+//
+// Element comparisons reuse DataTypeCollisionHandledFieldName so a numeric literal against an
+// Int64 array (or a numeric literal against a String array) no longer silently misses.
+func (c *jsonConditionBuilder) buildArrayFunctionCondition(operator qbtypes.FilterOperator, value any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	if len(c.key.JSONPlan) == 0 {
+		return "", errors.NewInvalidInputf(errors.CodeInvalidInput, "function `%s` could not resolve a JSON access plan for field `%s`", operator.FunctionName(), c.key.Name)
+	}
+
+	switch operator {
+	case qbtypes.FilterOperatorHas, qbtypes.FilterOperatorHasAny:
+		return c.buildOredRootChains(func(node *telemetrytypes.JSONAccessNode) (string, error) {
+			return c.arrayFunctionLeaf(node, operator, value, sb)
+		}, sb)
+	case qbtypes.FilterOperatorHasAll:
+		// contains-all: AND of a per-value "has" so the AND sits outside the array hops.
+		values := toAnyList(value)
+		conditions := make([]string, 0, len(values))
+		for _, v := range values {
+			v := v
+			cond, err := c.buildOredRootChains(func(node *telemetrytypes.JSONAccessNode) (string, error) {
+				return c.arrayFunctionLeaf(node, qbtypes.FilterOperatorHas, v, sb)
+			}, sb)
+			if err != nil {
+				return "", err
+			}
+			conditions = append(conditions, cond)
+		}
+		if len(conditions) == 1 {
+			return conditions[0], nil
+		}
+		return sb.And(conditions...), nil
+	}
+	return "", qbtypes.ErrUnsupportedOperator
+}
+
+// buildOredRootChains applies leafFn down every JSONPlan root (base + promoted), wrapping each
+// in its arrayExists chain, and ORs the per-root results.
+func (c *jsonConditionBuilder) buildOredRootChains(leafFn func(*telemetrytypes.JSONAccessNode) (string, error), sb *sqlbuilder.SelectBuilder) (string, error) {
+	conditions := make([]string, 0, len(c.key.JSONPlan))
+	for _, root := range c.key.JSONPlan {
+		cond, err := c.buildArrayExistsChain(root, leafFn, sb)
+		if err != nil {
+			return "", err
+		}
+		conditions = append(conditions, cond)
+	}
+	if len(conditions) == 1 {
+		return conditions[0], nil
+	}
+	return sb.Or(conditions...), nil
+}
+
+// buildArrayExistsChain wraps the terminal condition (produced by leafFn) in an arrayExists
+// over every array hop between the root and the terminal. For a terminal root (a top-level
+// array leaf) it simply returns leafFn(root).
+func (c *jsonConditionBuilder) buildArrayExistsChain(node *telemetrytypes.JSONAccessNode, leafFn func(*telemetrytypes.JSONAccessNode) (string, error), sb *sqlbuilder.SelectBuilder) (string, error) {
+	if node == nil {
+		return "", errors.NewInternalf(CodeArrayNavigationFailed, "navigation failed, current node is nil")
+	}
+	if node.IsTerminal {
+		return leafFn(node)
+	}
+
+	branches := make([]string, 0, 2)
+	for _, branch := range node.BranchesInOrder() {
+		childCond, err := c.buildArrayExistsChain(node.Branches[branch], leafFn, sb)
+		if err != nil {
+			return "", err
+		}
+		branches = append(branches, fmt.Sprintf("arrayExists(%s-> %s, %s)", node.Alias(), childCond, c.branchArrayExpr(node, branch)))
+	}
+	if len(branches) == 1 {
+		return branches[0], nil
+	}
+	return sb.Or(branches...), nil
+}
+
+// arrayFunctionLeaf builds the existential comparison for has/hasAny at a terminal node (hasAll
+// composes from has in buildArrayFunctionCondition). For an array leaf it delegates to native
+// membership; for a scalar leaf it compares the element directly.
+func (c *jsonConditionBuilder) arrayFunctionLeaf(node *telemetrytypes.JSONAccessNode, operator qbtypes.FilterOperator, value any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	if node.TerminalConfig.ElemType.IsArray {
+		return c.arrayLeafMembership(node, operator, value, sb)
+	}
+
+	switch operator {
+	case qbtypes.FilterOperatorHas:
+		return c.arrayFuncScalarLeaf(node, qbtypes.FilterOperatorEqual, value, sb)
+	case qbtypes.FilterOperatorHasAny:
+		return c.arrayFuncScalarLeaf(node, qbtypes.FilterOperatorIn, toAnyList(value), sb)
+	}
+	return "", qbtypes.ErrUnsupportedOperator
+}
+
+// arrayLeafMembership builds native membership for an array-typed leaf, reusing
+// buildArrayMembershipCondition (which handles data-type collisions on each element).
+func (c *jsonConditionBuilder) arrayLeafMembership(node *telemetrytypes.JSONAccessNode, operator qbtypes.FilterOperator, value any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	switch operator {
+	case qbtypes.FilterOperatorHas:
+		return c.buildArrayMembershipCondition(node, qbtypes.FilterOperatorEqual, value, sb)
+	case qbtypes.FilterOperatorHasAny:
+		return c.buildArrayMembershipCondition(node, qbtypes.FilterOperatorIn, toAnyList(value), sb)
+	}
+	return "", qbtypes.ErrUnsupportedOperator
+}
+
+// arrayFuncScalarLeaf builds `<elemExpr> <op> value` for a scalar leaf reached through an
+// array hop, applying data-type collision handling like the standard primitive path.
+func (c *jsonConditionBuilder) arrayFuncScalarLeaf(node *telemetrytypes.JSONAccessNode, operator qbtypes.FilterOperator, value any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	fieldExpr := fmt.Sprintf("dynamicElement(%s, '%s')", node.FieldPath(), node.TerminalConfig.ElemType.StringValue())
+	fieldExpr, value = querybuilder.DataTypeCollisionHandledFieldName(node.TerminalConfig.Key, value, fieldExpr, operator)
+	return c.applyOperator(sb, fieldExpr, operator, value)
+}
+
+// buildTokenFunctionCondition builds a hasToken search over a body JSON string field:
+// hasToken(LOWER(<elem>), LOWER(?)) wrapped in arrayExists over any array hops between the
+// root and the terminal. The field must resolve to a String leaf or a String array.
+func (c *jsonConditionBuilder) buildTokenFunctionCondition(needle any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	if len(c.key.JSONPlan) == 0 {
+		return "", errors.NewInvalidInputf(errors.CodeInvalidInput, "function `hasToken` could not resolve a JSON access plan for field `%s`", c.key.Name)
+	}
+
+	return c.buildOredRootChains(func(node *telemetrytypes.JSONAccessNode) (string, error) {
+		return c.tokenLeaf(node, needle, sb)
+	}, sb)
+}
+
+// tokenLeaf builds the hasToken match at a terminal node: a direct match for a String leaf,
+// or an arrayExists over the elements for a String array leaf. hasToken is string-only, so
+// any other element type is rejected.
+func (c *jsonConditionBuilder) tokenLeaf(node *telemetrytypes.JSONAccessNode, needle any, sb *sqlbuilder.SelectBuilder) (string, error) {
+	switch node.TerminalConfig.ElemType {
+	case telemetrytypes.String:
+		fieldExpr := fmt.Sprintf("dynamicElement(%s, 'String')", node.FieldPath())
+		return fmt.Sprintf("hasToken(LOWER(%s), LOWER(%s))", fieldExpr, sb.Var(needle)), nil
+	case telemetrytypes.ArrayString:
+		arrayExpr := fmt.Sprintf("dynamicElement(%s, '%s')", node.FieldPath(), node.TerminalConfig.ElemType.StringValue())
+		return fmt.Sprintf("arrayExists(x -> hasToken(LOWER(x), LOWER(%s)), %s)", sb.Var(needle), arrayExpr), nil
+	default:
+		return "", errors.NewInvalidInputf(errors.CodeInvalidInput, "function `hasToken` only supports string fields; field `%s` is `%s`", c.key.Name, node.TerminalConfig.Key.FieldDataType.StringValue())
+	}
+}
+
+// toAnyList normalizes a has-family value into a slice; a scalar becomes a one-element list.
+func toAnyList(value any) []any {
+	if list, ok := value.([]any); ok {
+		return list
+	}
+	return []any{value}
+}
+
 func (c *jsonConditionBuilder) applyOperator(sb *sqlbuilder.SelectBuilder, fieldExpr string, operator qbtypes.FilterOperator, value any) (string, error) {
 	switch operator {
 	case qbtypes.FilterOperatorEqual:

pkg/telemetrylogs/json_stmt_builder_test.go

@@ -602,7 +602,7 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			name:   "Simple has filter",
 			filter: "has(body.education[].parameters, 1.65)",
 			expected: TestExpected{
-				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))')))), ?))",
+				WhereClause: "(arrayExists(`body_v2.education`-> 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))')) OR arrayExists(`body_v2.education`-> 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{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].",
@@ -613,8 +613,8 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			name:   "Flat path hasAll filter",
 			filter: "hasAll(body.user.permissions, ['read', 'write'])",
 			expected: TestExpected{
-				WhereClause: "hasAll(dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))'), ?)",
-				Args:        []any{[]any{"read", "write"}, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+				WhereClause: "(arrayExists(x -> x = ?, dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))')) AND arrayExists(x -> x = ?, dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))')))",
+				Args:        []any{"read", "write", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 			},
 		},
 
@@ -739,8 +739,8 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			name:   "Nested path hasAny filter",
 			filter: "hasAny(education[].awards[].participated[].members, ['Piyush', 'Tushar'])",
 			expected: TestExpected{
-				WhereClause: "hasAny(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->arrayConcat(arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?)",
-				Args:        []any{[]any{"Piyush", "Tushar"}, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+				WhereClause: "arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> x IN (?, ?), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> x IN (?, ?), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> x IN (?, ?), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> x IN (?, ?), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"Piyush", "Tushar", "Piyush", "Tushar", "Piyush", "Tushar", "Piyush", "Tushar", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 			},
 		},
 		{
@@ -755,8 +755,8 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			name:   "dynamic_array_element_compare_HAS_STRING",
 			filter: "has(interests[].entities[].product_codes, '2002')",
 			expected: TestExpected{
-				WhereClause: "has(arrayFlatten(arrayConcat(arrayMap(`body_v2.interests`->arrayMap(`body_v2.interests[].entities`->dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'), 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))')))), ?)",
-				Args:        []any{"2002", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+				WhereClause: "arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'))), 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))'))",
+				Args:        []any{int64(2002), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 			},
 		},
 		{
@@ -771,10 +771,126 @@ func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
 			name:   "dynamic_array_element_compare_HAS_INT",
 			filter: "has(interests[].entities[].product_codes, 1001)",
 			expected: TestExpected{
-				WhereClause: "has(arrayFlatten(arrayConcat(arrayMap(`body_v2.interests`->arrayMap(`body_v2.interests[].entities`->dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'), 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))')))), ?)",
+				WhereClause: "arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'))), 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))'))",
 				Args:        []any{float64(1001), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 			},
 		},
+
+		// ── scalar leaf reached through an array ───
+		{
+			name:   "Nested primitive leaf has",
+			filter: "has(body.education[].name, 'IIT')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"IIT", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			name:   "Nested primitive leaf hasAny",
+			filter: "hasAny(body.education[].name, ['IIT', 'MIT'])",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') IN (?, ?), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"IIT", "MIT", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			// contains-all: single-value hasAll over a nested leaf collapses to has.
+			name:   "Nested primitive leaf hasAll single collapses to has",
+			filter: "hasAll(body.education[].name, 'IIT')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"IIT", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			// contains-all: "some element is IIT AND some element is MIT" (AND of per-value has).
+			name:   "Nested primitive leaf hasAll multi (contains-all)",
+			filter: "hasAll(body.education[].name, ['IIT', 'MIT'])",
+			expected: TestExpected{
+				WhereClause: "(arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')) AND arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
+				Args:        []any{"IIT", "MIT", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		// ── numeric literal against an Int64 array is collision-handled ───
+		{
+			name:   "Nested Int64 array has collision",
+			filter: "has(body.education[].scores, 90)",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.education`.`scores`, 'Array(Nullable(Int64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{float64(90), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		// ── hasAny folds multiple scalar arguments into one value set ─────
+		{
+			name:   "hasAny folds multiple scalar args",
+			filter: "hasAny(body.user.permissions, 'read', 'write')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(x -> x IN (?, ?), dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))'))",
+				Args:        []any{"read", "write", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		// ── hasToken over JSON string fields (nested leaf, top-level array, nested array) ──
+		{
+			name:   "hasToken nested string leaf",
+			filter: "hasToken(body.education[].name, 'harvard')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> hasToken(LOWER(dynamicElement(`body_v2.education`.`name`, 'String')), LOWER(?)), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"harvard", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			name:   "hasToken top-level string array",
+			filter: "hasToken(body.user.permissions, 'admin')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(x -> hasToken(LOWER(x), LOWER(?)), dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))'))",
+				Args:        []any{"admin", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			name:   "hasToken nested string array",
+			filter: "hasToken(body.education[].awards[].participated[].members, 'piyush')",
+			expected: TestExpected{
+				WhereClause: "arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> hasToken(LOWER(x), LOWER(?)), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> hasToken(LOWER(x), LOWER(?)), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> hasToken(LOWER(x), LOWER(?)), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(x -> hasToken(LOWER(x), LOWER(?)), dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))",
+				Args:        []any{"piyush", "piyush", "piyush", "piyush", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+
+		// ── scalar (non-array) leaf: treated as a single-element set ─────────────
+		{
+			name:   "Scalar leaf has",
+			filter: "has(body.user.name, 'alice')",
+			expected: TestExpected{
+				WhereClause: "dynamicElement(body_v2.`user.name`, 'String') = ?",
+				Args:        []any{"alice", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			name:   "Scalar leaf hasAny",
+			filter: "hasAny(body.user.name, ['alice', 'bob'])",
+			expected: TestExpected{
+				WhereClause: "dynamicElement(body_v2.`user.name`, 'String') IN (?, ?)",
+				Args:        []any{"alice", "bob", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			// contains-all: single-value hasAll over a plain scalar collapses to has.
+			name:   "Scalar leaf hasAll single collapses to has",
+			filter: "hasAll(body.user.name, 'alice')",
+			expected: TestExpected{
+				WhereClause: "dynamicElement(body_v2.`user.name`, 'String') = ?",
+				Args:        []any{"alice", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
+		{
+			// contains-all over a one-element set: the scalar must equal every value, so
+			// distinct values never match.
+			name:   "Scalar leaf hasAll multi (contains-all)",
+			filter: "hasAll(body.user.name, ['alice', 'bob'])",
+			expected: TestExpected{
+				WhereClause: "(dynamicElement(body_v2.`user.name`, 'String') = ? AND dynamicElement(body_v2.`user.name`, 'String') = ?)",
+				Args:        []any{"alice", "bob", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+			},
+		},
 	}
 
 	for _, c := range cases {

pkg/telemetrylogs/json_string.go

@@ -130,3 +130,121 @@ func GetBodyJSONKey(_ context.Context, key *telemetrytypes.TelemetryFieldKey, op
 func GetBodyJSONKeyForExists(_ context.Context, key *telemetrytypes.TelemetryFieldKey, _ qbtypes.FilterOperator, _ any) string {
 	return fmt.Sprintf("JSON_EXISTS(body, '$.%s')", getBodyJSONPath(key))
 }
+
+// legacyElemType infers the has-family element type from the needle (legacy has no schema). It
+// scans EVERY value so the chosen array type and all coerced needles agree — else ClickHouse
+// raises "no supertype ... String" (code 386). Int64 stays distinct from Float64 so a quoted
+// integer is exact past 2^53 (unquoted literals already arrive as float64, parsed upstream).
+func legacyElemType(needle any) telemetrytypes.FieldDataType {
+	list, ok := needle.([]any)
+	if !ok {
+		list = []any{needle}
+	}
+	if len(list) == 0 {
+		return telemetrytypes.FieldDataTypeString
+	}
+	allInt, allNumeric := true, true
+	for _, v := range list {
+		switch t := v.(type) {
+		case float32, float64:
+			allInt = false
+		case int, int8, int16, int32, int64, uint, uint8, uint16, uint32, uint64:
+			// integer Go types stay int-exact
+		case string:
+			if _, err := strconv.ParseInt(t, 10, 64); err != nil {
+				allInt = false
+			}
+			if _, err := strconv.ParseFloat(t, 64); err != nil {
+				allNumeric = false
+			}
+		default:
+			// booleans (and anything else) -> String; a bool renders to 'true'/'false', so a
+			// bool needle only matches genuine JSON booleans, not truthy numbers/strings.
+			allInt, allNumeric = false, false
+		}
+	}
+	switch {
+	case allInt:
+		return telemetrytypes.FieldDataTypeInt64
+	case allNumeric:
+		return telemetrytypes.FieldDataTypeFloat64
+	default:
+		return telemetrytypes.FieldDataTypeString
+	}
+}
+
+// legacyCoerceNeedle coerces a needle to elem type dt so its bound-arg type matches the
+// extracted column (legacyElemType guarantees it's coercible).
+func legacyCoerceNeedle(v any, dt telemetrytypes.FieldDataType) any {
+	switch dt {
+	case telemetrytypes.FieldDataTypeInt64:
+		if s, ok := v.(string); ok {
+			if i, err := strconv.ParseInt(s, 10, 64); err == nil {
+				return i
+			}
+		}
+		return v
+	case telemetrytypes.FieldDataTypeFloat64:
+		if s, ok := v.(string); ok {
+			f, _ := strconv.ParseFloat(s, 64)
+			return f
+		}
+		return v
+	default:
+		return bodyArrayNeedleString(v)
+	}
+}
+
+// getBodyJSONArrayKey extracts the leaf as Array(Nullable(<dt>)) — Nullable so a value of a
+// different JSON type maps to NULL instead of corrupting (e.g. a non-numeric string → 0).
+func getBodyJSONArrayKey(key *telemetrytypes.TelemetryFieldKey, dt telemetrytypes.FieldDataType) string {
+	arrKey := *key
+	if !strings.HasSuffix(arrKey.Name, "[*]") && !strings.HasSuffix(arrKey.Name, "[]") {
+		arrKey.Name += "[*]"
+	}
+	return fmt.Sprintf("JSONExtract(JSON_QUERY(body, '$.%s'), 'Array(Nullable(%s))')", getBodyJSONPath(&arrKey), dt.CHDataType())
+}
+
+// getBodyJSONScalarKey builds the single-element-set fallback for a scalar body value: the leaf
+// extracted as a scalar of type dt, plus a guard restricting it to a genuinely scalar body. The
+// guard is required because JSON_VALUE returns '' for an array/object/missing value, which would
+// otherwise zero-value match (has(x,0) / has(x,false) / has(x,'') on any array). ok=false when
+// the path still traverses an array ([*]/[]).
+func getBodyJSONScalarKey(key *telemetrytypes.TelemetryFieldKey, dt telemetrytypes.FieldDataType) (expr string, guard string, ok bool) {
+	name := strings.TrimSuffix(strings.TrimSuffix(key.Name, "[*]"), "[]")
+	if strings.Contains(name, "[") {
+		return "", "", false
+	}
+	scalarKey := *key
+	scalarKey.Name = name
+	path := getBodyJSONPath(&scalarKey)
+	if dt == telemetrytypes.FieldDataTypeString {
+		expr = fmt.Sprintf("JSON_VALUE(body, '$.%s')", path)
+	} else {
+		// Nullable so a scalar of a different type (e.g. a bool/string where a number is
+		// searched) extracts to NULL rather than the type's default (0/false), which would
+		// otherwise zero-value match has(x, 0).
+		expr = fmt.Sprintf("JSONExtract(JSON_VALUE(body, '$.%s'), 'Nullable(%s)')", path, dt.CHDataType())
+	}
+	keys := strings.Split(name, ".")
+	for i, k := range keys {
+		keys[i] = "'" + k + "'"
+	}
+	guard = fmt.Sprintf("JSONType(body, %s) NOT IN ('Array', 'Object', 'Null')", strings.Join(keys, ", "))
+	return expr, guard, true
+}
+
+func bodyArrayNeedleString(v any) string {
+	switch t := v.(type) {
+	case string:
+		return t
+	case bool:
+		return strconv.FormatBool(t)
+	case float64:
+		return strconv.FormatFloat(t, 'f', -1, 64)
+	case float32:
+		return strconv.FormatFloat(float64(t), 'f', -1, 64)
+	default:
+		return fmt.Sprintf("%v", t)
+	}
+}

pkg/telemetrylogs/stmt_builder_test.go

@@ -495,8 +495,8 @@ func TestStatementBuilderListQueryResourceTests(t *testing.T) {
 				Limit: 10,
 			},
 			expected: qbtypes.Statement{
-				Query: "SELECT timestamp, id, trace_id, span_id, trace_flags, severity_text, severity_number, scope_name, scope_version, body, attributes_string, attributes_number, attributes_bool, resources_string, scope_string FROM signoz_logs.distributed_logs_v2 WHERE has(JSONExtract(JSON_QUERY(body, '$.\"user_names\"[*]'), 'Array(String)'), ?) AND timestamp >= ? AND ts_bucket_start >= ? AND timestamp < ? AND ts_bucket_start <= ? LIMIT ?",
-				Args:  []any{"john_doe", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
+				Query: "SELECT timestamp, id, trace_id, span_id, trace_flags, severity_text, severity_number, scope_name, scope_version, body, attributes_string, attributes_number, attributes_bool, resources_string, scope_string FROM signoz_logs.distributed_logs_v2 WHERE (has(JSONExtract(JSON_QUERY(body, '$.\"user_names\"[*]'), 'Array(Nullable(String))'), ?) OR (JSON_VALUE(body, '$.\"user_names\"') = ? AND JSONType(body, 'user_names') NOT IN ('Array', 'Object', 'Null'))) AND timestamp >= ? AND ts_bucket_start >= ? AND timestamp < ? AND ts_bucket_start <= ? LIMIT ?",
+				Args:  []any{"john_doe", "john_doe", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
 			},
 			expectedErr: nil,
 		},

tests/integration/tests/querier_json_body/03_body_array_functions.py

@@ -3,17 +3,18 @@ from collections.abc import Callable
 from datetime import UTC, datetime, timedelta
 from http import HTTPStatus
 
+import pytest
+
 from fixtures import types
 from fixtures.auth import USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD
 from fixtures.logs import Logs
-from fixtures.querier import get_rows, make_query_request
-
-# Positive coverage for the body-JSON array functions hasAny / hasAll in JSON-body
-# mode (BODY_JSON_QUERY_ENABLED=true). Here body_v2 arrays resolve to real ClickHouse
-# Arrays via dynamicElement, so these succeed — unlike legacy body mode, where
-# hasAny/hasAll xfail (see querierlogs/09_json_body_functions.py).
-# export_json_types registers the body paths + array element types (tags -> []string,
-# ids -> []int64) so the builder resolves body.tags/body.ids as arrays.
+from fixtures.querier import (
+    RequestType,
+    build_order_by,
+    build_raw_query,
+    get_rows,
+    make_query_request,
+)
 
 
 def test_logs_json_body_has_any_string(
@@ -23,7 +24,7 @@ def test_logs_json_body_has_any_string(
     insert_logs: Callable[[list[Logs]], None],
     export_json_types: Callable[[list[Logs]], None],
 ) -> None:
-    """hasAny over a []string body array: matches logs sharing ANY listed value."""
+    """hasAny over a []string body array matches logs sharing any listed value."""
     now = datetime.now(tz=UTC)
     logs = [
         Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body_v2=json.dumps({"tags": ["production", "api", "critical"]}), body_promoted="", severity_text="INFO"),
@@ -34,16 +35,15 @@ def test_logs_json_body_has_any_string(
     insert_logs(logs)
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # log1 has "critical", log2 has "test", log3 has neither -> 2 matches
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
-        request_type="raw",
-        queries=[{"type": "builder_query", "spec": {"name": "A", "signal": "logs", "disabled": False, "limit": 100, "offset": 0, "filter": {"expression": "hasAny(body.tags, ['critical', 'test'])"}, "order": [{"key": {"name": "timestamp"}, "direction": "desc"}], "aggregations": [{"expression": "count()"}]}}],
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAny(body.tags, ['critical', 'test'])")],
     )
-    assert response.status_code == HTTPStatus.OK
+    assert response.status_code == HTTPStatus.OK, response.text
     rows = get_rows(response)
     assert len(rows) == 2
     assert all(("critical" in row["data"]["body"]["tags"]) or ("test" in row["data"]["body"]["tags"]) for row in rows)
@@ -56,7 +56,7 @@ def test_logs_json_body_has_all_string(
     insert_logs: Callable[[list[Logs]], None],
     export_json_types: Callable[[list[Logs]], None],
 ) -> None:
-    """hasAll over a []string body array: matches only logs having ALL listed values."""
+    """hasAll over a []string body array matches only logs having all listed values."""
     now = datetime.now(tz=UTC)
     logs = [
         Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"tags": ["production", "api", "critical"]}), body_promoted="", severity_text="INFO"),
@@ -66,16 +66,15 @@ def test_logs_json_body_has_all_string(
     insert_logs(logs)
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # only the second log has both "production" AND "web"
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
-        request_type="raw",
-        queries=[{"type": "builder_query", "spec": {"name": "A", "signal": "logs", "disabled": False, "limit": 100, "offset": 0, "filter": {"expression": "hasAll(body.tags, ['production', 'web'])"}, "order": [{"key": {"name": "timestamp"}, "direction": "desc"}], "aggregations": [{"expression": "count()"}]}}],
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAll(body.tags, ['production', 'web'])")],
     )
-    assert response.status_code == HTTPStatus.OK
+    assert response.status_code == HTTPStatus.OK, response.text
     rows = get_rows(response)
     assert len(rows) == 1
     tags = rows[0]["data"]["body"]["tags"]
@@ -99,16 +98,382 @@ def test_logs_json_body_has_any_number(
     insert_logs(logs)
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # only the first log has 300 in ids; 999 matches nothing
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
-        request_type="raw",
-        queries=[{"type": "builder_query", "spec": {"name": "A", "signal": "logs", "disabled": False, "limit": 100, "offset": 0, "filter": {"expression": "hasAny(body.ids, [300, 999])"}, "order": [{"key": {"name": "timestamp"}, "direction": "desc"}], "aggregations": [{"expression": "count()"}]}}],
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAny(body.ids, [300, 999])")],
     )
-    assert response.status_code == HTTPStatus.OK
+    assert response.status_code == HTTPStatus.OK, response.text
     rows = get_rows(response)
     assert len(rows) == 1
     assert 300 in rows[0]["data"]["body"]["ids"]
+
+
+def test_logs_json_body_array_membership(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """has(body.<array>, value) matches logs whose body array contains the value, across
+    string, numeric and boolean element types."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body_v2=json.dumps({"tags": ["production", "api", "critical"], "ids": [100, 200, 300], "flags": [True, False, True]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"tags": ["staging", "api", "test"], "ids": [200, 400, 500], "flags": [False, False, True]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"tags": ["production", "web", "important"], "ids": [100, 600, 700], "flags": [True, True, False]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='has(body.tags, "production")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 2
+    assert all("production" in row["data"]["body"]["tags"] for row in rows)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.ids, 200)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 2
+    assert all(200 in row["data"]["body"]["ids"] for row in rows)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.flags, true)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 3
+    assert all(True in row["data"]["body"]["flags"] for row in rows)
+
+
+def test_logs_json_body_has_token(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """hasToken(body, token) matches logs whose body message contains the token."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"message": "request served from production node", "tags": ["api"]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"message": "request served from staging node", "tags": ["api"]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='hasToken(body, "production")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 1
+    assert "production" in rows[0]["data"]["body"]["message"]
+
+
+def test_logs_json_body_function_scalar_path(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """has/hasAny/hasAll on a scalar (non-array) body path treat the scalar as a single-element
+    set: has = equals, hasAny/hasAll = membership. So has and hasAll coincide for a scalar."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"level": "info", "code": 200}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"level": "debug", "code": 500}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    # Each expression matches only the first log (level=info, code=200).
+    for expression in ["has(body.level, 'info')", "hasAny(body.code, [200, 999])", "hasAll(body.level, ['info'])"]:
+        response = make_query_request(
+            signoz,
+            token,
+            start_ms=start_ms,
+            end_ms=end_ms,
+            request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: expected 200, got {response.status_code}: {response.text}"
+        rows = get_rows(response)
+        assert len(rows) == 1, f"{expression}: expected 1 match, got {len(rows)}"
+        assert rows[0]["data"]["body"]["level"] == "info", f"{expression}: matched wrong row: {rows[0]['data']['body']}"
+
+    # hasAll is contains-all: a scalar (single-element set) can't hold two distinct values, so nothing matches.
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAll(body.level, ['info', 'debug'])")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    assert get_rows(response) == [], "hasAll(scalar, [two distinct values]) should match nothing (contains-all)"
+
+
+@pytest.mark.parametrize(
+    "expression",
+    [
+        pytest.param('has(code.function, "main")', id="has_non_body_key"),
+        pytest.param('hasToken(code.function, "main")', id="hastoken_non_body_key"),
+        pytest.param("hasToken(body, 123)", id="hastoken_non_string_value"),
+    ],
+)
+def test_logs_json_body_function_errors(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    expression: str,
+) -> None:
+    """has/hasToken misuse (non-body key, non-string token) is rejected (400)."""
+    now = datetime.now(tz=UTC)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+    )
+    assert response.status_code == HTTPStatus.BAD_REQUEST
+
+
+@pytest.mark.parametrize(
+    "expression, expected_error",
+    [
+        pytest.param('has(body.tags, "a", "b")', "expects exactly one value argument", id="has_multiple_values"),
+        pytest.param('hasToken(body, "a", "b")', "expects exactly one value argument", id="hastoken_multiple_values"),
+        pytest.param('has(body.tags, ["a", "b"])', "expects a single scalar value, not an array", id="has_array_value"),
+        pytest.param('hasAny(body.tags, ["a"], "b")', "not a mix of the two", id="hasany_mixed_array_and_scalar"),
+    ],
+)
+def test_logs_json_body_function_argument_shape_errors(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    expression: str,
+    expected_error: str,
+) -> None:
+    """has-family value arguments are shape-validated; mis-shaped arguments are rejected (400)."""
+    now = datetime.now(tz=UTC)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+    )
+    assert response.status_code == HTTPStatus.BAD_REQUEST, f"{expression}: expected 400, got {response.status_code}: {response.text}"
+    assert expected_error in response.text, f"{expression}: expected error {expected_error!r} in {response.text}"
+
+
+
+def test_logs_json_body_large_integer_exact(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """Large ids (> 2^53) stay exact only when QUOTED (compared as Int64). An unquoted numeric
+    literal is parsed as float64 upstream, so it false-matches neighbouring ids (documented)."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"id": [1234567890123456789]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"id": [1234567890123456788]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    # Quoted -> Int64 -> exact: matches only the exact id.
+    response = make_query_request(
+        signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='has(body.id, "1234567890123456789")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    # exactly one match (the query compares as Int64); the returned body value itself is
+    # float64-serialized in the response, so assert on the match count, not the echoed id.
+    assert len(get_rows(response)) == 1, "quoted big id should match exactly one log"
+
+    # Unquoted -> float64 upstream -> imprecise: false-matches the off-by-one neighbour too.
+    response = make_query_request(
+        signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.id, 1234567890123456789)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    assert len(get_rows(response)) == 2, "unquoted big id collides with neighbour (float64 precision) - quote large ids"
+
+
+def test_logs_json_body_no_supertype_safety(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """A needle whose type doesn't match the body array returns 200 with no match (never a
+    ClickHouse type error)."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"names": ["alice", "bob"]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"nums": [100, 200]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression in ["has(body.names, 200)", "has(body.nums, 'alice')"]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: expected 200, got {response.status_code}: {response.text}"
+        assert get_rows(response) == [], f"{expression}: type mismatch should match nothing"
+
+
+def test_logs_json_body_type_isolation(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """JSON-body mode is inherently free of the legacy coercion quirks (it compares typed
+    dynamicElement values off body_v2): a bool needle matches only genuine booleans, and
+    empty-string / zero needles never false-match array bodies. Genuine scalars still match."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=4), resources={"service.name": "app-service"}, body_v2=json.dumps({"nums": [100, 200, 300]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body_v2=json.dumps({"flags": [True, False]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"names": ["a", "b"]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"code": 0}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [
+        ("has(body.nums, true)", 0),  # bool needle does NOT truthy-match a numeric array
+        ("has(body.flags, true)", 1),  # genuine bool array matches
+        ("has(body.names, '')", 0),  # empty-string does NOT false-match an array
+        ("has(body.nums, 0)", 0),  # zero does NOT false-match an array without a 0
+        ("has(body.code, 0)", 1),  # genuine scalar zero matches
+    ]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"
+
+
+def test_logs_json_body_dynamic_array(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """has() over a mixed-type body array (stored as Array(Dynamic)) matches by the needle's
+    type: a numeric needle matches numeric elements, a string needle matches string elements."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"mix": [100, "abc", 300]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [
+        ("has(body.mix, 100)", 1),
+        ("has(body.mix, 'abc')", 1),
+        ("has(body.mix, 999)", 0),
+        ("has(body.mix, 'xyz')", 0),
+    ]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"
+
+
+def test_logs_json_body_nested_array_path(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+    export_json_types: Callable[[list[Logs]], None],
+) -> None:
+    """has over a scalar leaf reached through an array (body.education[].name) checks each element."""
+    now = datetime.now(tz=UTC)
+    logs = [
+        Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body_v2=json.dumps({"education": [{"name": "IIT"}, {"name": "MIT"}]}), body_promoted="", severity_text="INFO"),
+        Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body_v2=json.dumps({"education": [{"name": "Stanford"}]}), body_promoted="", severity_text="INFO"),
+    ]
+    export_json_types(logs)
+    insert_logs(logs)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [("has(body.education[].name, 'MIT')", 1), ("has(body.education[].name, 'Yale')", 0)]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"

tests/integration/tests/querierlogs/06_json_body.py

@@ -563,214 +563,6 @@ def test_logs_json_body_nested_keys(
     assert all(code == 200 for code in status_codes)
 
 
-def test_logs_json_body_array_membership(
-    signoz: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_logs: Callable[[list[Logs]], None],
-) -> None:
-    """
-    Setup:
-    Insert logs with JSON bodies containing arrays
-
-    Tests:
-    1. Search by has(body.tags[*], "value") - string array
-    2. Search by has(body.ids[*], 123) - numeric array
-    3. Search by has(body.flags[*], true) - boolean array
-    """
-    now = datetime.now(tz=UTC)
-
-    log1_body = json.dumps(
-        {
-            "tags": ["production", "api", "critical"],
-            "ids": [100, 200, 300],
-            "flags": [True, False, True],
-            "users": [
-                {"name": "alice", "role": "admin"},
-                {"name": "bob", "role": "user"},
-            ],
-        }
-    )
-
-    log2_body = json.dumps(
-        {
-            "tags": ["staging", "api", "test"],
-            "ids": [200, 400, 500],
-            "flags": [False, False, True],
-            "users": [
-                {"name": "charlie", "role": "user"},
-                {"name": "david", "role": "admin"},
-            ],
-        }
-    )
-
-    log3_body = json.dumps(
-        {
-            "tags": ["production", "web", "important"],
-            "ids": [100, 600, 700],
-            "flags": [True, True, False],
-            "users": [
-                {"name": "alice", "role": "admin"},
-                {"name": "eve", "role": "user"},
-            ],
-        }
-    )
-
-    insert_logs(
-        [
-            Logs(
-                timestamp=now - timedelta(seconds=3),
-                resources={"service.name": "app-service"},
-                attributes={},
-                body=log1_body,
-                severity_text="INFO",
-            ),
-            Logs(
-                timestamp=now - timedelta(seconds=2),
-                resources={"service.name": "app-service"},
-                attributes={},
-                body=log2_body,
-                severity_text="INFO",
-            ),
-            Logs(
-                timestamp=now - timedelta(seconds=1),
-                resources={"service.name": "app-service"},
-                attributes={},
-                body=log3_body,
-                severity_text="INFO",
-            ),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-
-    # Test 1: Search by has(body.tags[*], "production")
-    response = requests.post(
-        signoz.self.host_configs["8080"].get("/api/v5/query_range"),
-        timeout=2,
-        headers={"authorization": f"Bearer {token}"},
-        json={
-            "schemaVersion": "v1",
-            "start": int((now - timedelta(seconds=10)).timestamp() * 1000),
-            "end": int(now.timestamp() * 1000),
-            "requestType": "raw",
-            "compositeQuery": {
-                "queries": [
-                    {
-                        "type": "builder_query",
-                        "spec": {
-                            "name": "A",
-                            "signal": "logs",
-                            "disabled": False,
-                            "limit": 100,
-                            "offset": 0,
-                            "filter": {"expression": 'has(body.tags[*], "production")'},
-                            "order": [
-                                {"key": {"name": "timestamp"}, "direction": "desc"},
-                            ],
-                            "aggregations": [{"expression": "count()"}],
-                        },
-                    }
-                ]
-            },
-            "formatOptions": {"formatTableResultForUI": False, "fillGaps": False},
-        },
-    )
-
-    assert response.status_code == HTTPStatus.OK
-    assert response.json()["status"] == "success"
-    results = response.json()["data"]["data"]["results"]
-    assert len(results) == 1
-    rows = results[0]["rows"]
-    assert len(rows) == 2  # log1 and log3 have "production" in tags
-    tags_list = [json.loads(row["data"]["body"])["tags"] for row in rows]
-    assert all("production" in tags for tags in tags_list)
-
-    # Test 2: Search by has(body.ids[*], 200)
-    response = requests.post(
-        signoz.self.host_configs["8080"].get("/api/v5/query_range"),
-        timeout=2,
-        headers={"authorization": f"Bearer {token}"},
-        json={
-            "schemaVersion": "v1",
-            "start": int((now - timedelta(seconds=10)).timestamp() * 1000),
-            "end": int(now.timestamp() * 1000),
-            "requestType": "raw",
-            "compositeQuery": {
-                "queries": [
-                    {
-                        "type": "builder_query",
-                        "spec": {
-                            "name": "A",
-                            "signal": "logs",
-                            "disabled": False,
-                            "limit": 100,
-                            "offset": 0,
-                            "filter": {"expression": "has(body.ids[*], 200)"},
-                            "order": [
-                                {"key": {"name": "timestamp"}, "direction": "desc"},
-                            ],
-                            "aggregations": [{"expression": "count()"}],
-                        },
-                    }
-                ]
-            },
-            "formatOptions": {"formatTableResultForUI": False, "fillGaps": False},
-        },
-    )
-
-    assert response.status_code == HTTPStatus.OK
-    assert response.json()["status"] == "success"
-    results = response.json()["data"]["data"]["results"]
-    assert len(results) == 1
-    rows = results[0]["rows"]
-    assert len(rows) == 2  # log1 and log2 have 200 in ids
-    ids_list = [json.loads(row["data"]["body"])["ids"] for row in rows]
-    assert all(200 in ids for ids in ids_list)
-
-    # Test 3: Search by has(body.flags[*], true)
-    response = requests.post(
-        signoz.self.host_configs["8080"].get("/api/v5/query_range"),
-        timeout=2,
-        headers={"authorization": f"Bearer {token}"},
-        json={
-            "schemaVersion": "v1",
-            "start": int((now - timedelta(seconds=10)).timestamp() * 1000),
-            "end": int(now.timestamp() * 1000),
-            "requestType": "raw",
-            "compositeQuery": {
-                "queries": [
-                    {
-                        "type": "builder_query",
-                        "spec": {
-                            "name": "A",
-                            "signal": "logs",
-                            "disabled": False,
-                            "limit": 100,
-                            "offset": 0,
-                            "filter": {"expression": "has(body.flags[*], true)"},
-                            "order": [
-                                {"key": {"name": "timestamp"}, "direction": "desc"},
-                            ],
-                            "aggregations": [{"expression": "count()"}],
-                        },
-                    }
-                ]
-            },
-            "formatOptions": {"formatTableResultForUI": False, "fillGaps": False},
-        },
-    )
-
-    assert response.status_code == HTTPStatus.OK
-    assert response.json()["status"] == "success"
-    results = response.json()["data"]["data"]["results"]
-    assert len(results) == 1
-    rows = results[0]["rows"]
-    assert len(rows) == 3  # All logs have true in flags
-    flags_list = [json.loads(row["data"]["body"])["flags"] for row in rows]
-    assert all(True in flags for flags in flags_list)
-
-
 def test_logs_json_body_listing(
     signoz: types.SigNoz,
     create_user_admin: None,  # pylint: disable=unused-argument

tests/integration/tests/querierlogs/09_json_body_functions.py

@@ -16,10 +16,161 @@ from fixtures.querier import (
     make_query_request,
 )
 
-# Body-JSON array/token functions on the logs body. has() success paths are already
-# covered by 06_json_body.py::test_logs_json_body_array_membership; this file adds the
-# sibling functions hasAny / hasAll / hasToken (success) and the function-operator error
-# paths (has/hasToken on a non-body key, non-string token) which must be rejected.
+
+def test_logs_json_body_array_membership(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """has(body.<array>[*], value) matches logs whose body array contains the value, across
+    string, numeric and boolean element types."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(
+                timestamp=now - timedelta(seconds=3),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"tags": ["production", "api", "critical"], "ids": [100, 200, 300], "flags": [True, False, True]}),
+                severity_text="INFO",
+            ),
+            Logs(
+                timestamp=now - timedelta(seconds=2),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"tags": ["staging", "api", "test"], "ids": [200, 400, 500], "flags": [False, False, True]}),
+                severity_text="INFO",
+            ),
+            Logs(
+                timestamp=now - timedelta(seconds=1),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"tags": ["production", "web", "important"], "ids": [100, 600, 700], "flags": [True, True, False]}),
+                severity_text="INFO",
+            ),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='has(body.tags[*], "production")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 2
+    assert all("production" in json.loads(row["data"]["body"])["tags"] for row in rows)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.ids[*], 200)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 2
+    assert all(200 in json.loads(row["data"]["body"])["ids"] for row in rows)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=start_ms,
+        end_ms=end_ms,
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.flags[*], true)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 3
+    assert all(True in json.loads(row["data"]["body"])["flags"] for row in rows)
+
+
+def test_logs_json_body_has_type_collision(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """A numeric-looking string needle matches a numeric body array (the needle is coerced to a number)."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(
+                timestamp=now - timedelta(seconds=2),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"ids": [100, 200, 300]}),
+                severity_text="INFO",
+            ),
+            Logs(
+                timestamp=now - timedelta(seconds=1),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"ids": [400, 500]}),
+                severity_text="INFO",
+            ),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='has(body.ids, "200")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 1
+    assert 200 in json.loads(rows[0]["data"]["body"])["ids"]
+
+
+def test_logs_json_body_has_mixed_type_array(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """has() over a mixed-type body array must not error (no ClickHouse "no supertype"). The
+    needle's type selects the view: a numeric needle matches numeric elements (non-numeric →
+    NULL, skipped), a string needle matches every element rendered as text."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(
+                timestamp=now - timedelta(seconds=1),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"ids": [100, "abc", 300]}),
+                severity_text="INFO",
+            ),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [
+        ("has(body.ids, 100)", 1),  # numeric needle → matches the numeric element
+        ("has(body.ids, 999)", 0),  # numeric needle → no match, non-numeric "abc" is NULL not 0
+        ("has(body.ids, 'abc')", 1),  # string needle → matches the string element
+    ]:
+        response = make_query_request(
+            signoz,
+            token,
+            start_ms=start_ms,
+            end_ms=end_ms,
+            request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: expected 200, got {response.status_code}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} match(es)"
 
 
 def test_logs_json_body_has_any(
@@ -28,7 +179,7 @@ def test_logs_json_body_has_any(
     get_token: Callable[[str, str], str],
     insert_logs: Callable[[list[Logs]], None],
 ) -> None:
-    """hasAny(body.tags, [...]) matches a log whose array shares ANY value."""
+    """hasAny(body.tags, [...]) matches a log whose array shares any listed value."""
     now = datetime.now(tz=UTC)
     insert_logs(
         [
@@ -54,26 +205,18 @@ def test_logs_json_body_has_any(
     )
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # log1 has "critical", log2 has "test", log3 has neither -> 2 matches
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
         request_type=RequestType.RAW,
-        queries=[
-            build_raw_query(
-                "A",
-                "logs",
-                order=[build_order_by("timestamp")],
-                limit=100,
-                filter_expression="hasAny(body.tags, ['critical', 'test'])",
-            )
-        ],
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAny(body.tags, ['critical', 'test'])")],
     )
-    # BUG: hasAny on a flat body array path extracts a scalar String; ClickHouse rejects
-    # it ("Argument 0 ... must be an array") -> HTTP 500.
-    assert response.status_code == HTTPStatus.INTERNAL_SERVER_ERROR
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 2
+    assert all(("critical" in json.loads(row["data"]["body"])["tags"]) or ("test" in json.loads(row["data"]["body"])["tags"]) for row in rows)
 
 
 def test_logs_json_body_has_all(
@@ -82,7 +225,7 @@ def test_logs_json_body_has_all(
     get_token: Callable[[str, str], str],
     insert_logs: Callable[[list[Logs]], None],
 ) -> None:
-    """hasAll(body.tags, [...]) matches only a log whose array has ALL values."""
+    """hasAll(body.tags, [...]) matches only a log whose array has all listed values."""
     now = datetime.now(tz=UTC)
     insert_logs(
         [
@@ -102,26 +245,19 @@ def test_logs_json_body_has_all(
     )
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # only the second log has both "production" AND "web"
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
         request_type=RequestType.RAW,
-        queries=[
-            build_raw_query(
-                "A",
-                "logs",
-                order=[build_order_by("timestamp")],
-                limit=100,
-                filter_expression="hasAll(body.tags, ['production', 'web'])",
-            )
-        ],
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAll(body.tags, ['production', 'web'])")],
     )
-    # BUG: hasAll on a flat body array path extracts a scalar String; ClickHouse rejects
-    # it ("Argument 0 ... must be an array") -> HTTP 500.
-    assert response.status_code == HTTPStatus.INTERNAL_SERVER_ERROR
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 1
+    tags = json.loads(rows[0]["data"]["body"])["tags"]
+    assert "production" in tags and "web" in tags
 
 
 def test_logs_json_body_has_token(
@@ -156,22 +292,13 @@ def test_logs_json_body_has_token(
     )
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    # "production" appears in the first and third log bodies
     response = make_query_request(
         signoz,
         token,
         start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
         end_ms=int(now.timestamp() * 1000),
         request_type=RequestType.RAW,
-        queries=[
-            build_raw_query(
-                "A",
-                "logs",
-                order=[build_order_by("timestamp")],
-                limit=100,
-                filter_expression='hasToken(body, "production")',
-            )
-        ],
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='hasToken(body, "production")')],
     )
     assert response.status_code == HTTPStatus.OK
     rows = get_rows(response)
@@ -179,28 +306,35 @@ def test_logs_json_body_has_token(
     assert all("production" in row["data"]["body"] for row in rows)
 
 
-def test_logs_json_body_function_scalar_path_errors(
+def test_logs_json_body_function_scalar_path(
     signoz: types.SigNoz,
     create_user_admin: None,  # pylint: disable=unused-argument
     get_token: Callable[[str, str], str],
     insert_logs: Callable[[list[Logs]], None],
 ) -> None:
-    """BUG: has/hasAny/hasAll on a scalar body path extract a scalar String; ClickHouse
-    rejects the array function -> HTTP 500."""
+    """has/hasAny/hasAll on a scalar (non-array) body path treat the scalar as a single-element
+    set: has = equals, hasAny/hasAll = membership. So has and hasAll coincide for a scalar."""
     now = datetime.now(tz=UTC)
     insert_logs(
         [
             Logs(
-                timestamp=now - timedelta(seconds=1),
+                timestamp=now - timedelta(seconds=2),
                 resources={"service.name": "app-service"},
                 body=json.dumps({"level": "info", "code": 200}),
                 severity_text="INFO",
             ),
+            Logs(
+                timestamp=now - timedelta(seconds=1),
+                resources={"service.name": "app-service"},
+                body=json.dumps({"level": "debug", "code": 500}),
+                severity_text="INFO",
+            ),
         ]
     )
     token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
 
-    for expression in ["has(body.level, 'info')", "hasAny(body.code, [200])", "hasAll(body.level, ['info'])"]:
+    # Each expression matches only the first log (level=info, code=200).
+    for expression in ["has(body.level, 'info')", "hasAny(body.code, [200, 999])", "hasAll(body.level, ['info'])"]:
         response = make_query_request(
             signoz,
             token,
@@ -209,7 +343,22 @@ def test_logs_json_body_function_scalar_path_errors(
             request_type=RequestType.RAW,
             queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
         )
-        assert response.status_code == HTTPStatus.INTERNAL_SERVER_ERROR, f"{expression}: expected 500, got {response.status_code}: {response.text}"
+        assert response.status_code == HTTPStatus.OK, f"{expression}: expected 200, got {response.status_code}: {response.text}"
+        rows = get_rows(response)
+        assert len(rows) == 1, f"{expression}: expected 1 match, got {len(rows)}"
+        assert json.loads(rows[0]["data"]["body"])["level"] == "info", f"{expression}: matched wrong row"
+
+    # hasAll is contains-all: a scalar (single-element set) can't hold two distinct values, so nothing matches.
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="hasAll(body.level, ['info', 'debug'])")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    assert get_rows(response) == [], "hasAll(scalar, [two distinct values]) should match nothing (contains-all)"
 
 
 @pytest.mark.parametrize(
@@ -238,3 +387,286 @@ def test_logs_json_body_function_errors(
         queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
     )
     assert response.status_code == HTTPStatus.BAD_REQUEST
+
+
+@pytest.mark.parametrize(
+    "expression, expected_error",
+    [
+        pytest.param('has(body.tags, "a", "b")', "expects exactly one value argument", id="has_multiple_values"),
+        pytest.param('hasToken(body, "a", "b")', "expects exactly one value argument", id="hastoken_multiple_values"),
+        pytest.param('has(body.tags, ["a", "b"])', "expects a single scalar value, not an array", id="has_array_value"),
+        pytest.param('hasAny(body.tags, ["a"], "b")', "not a mix of the two", id="hasany_mixed_array_and_scalar"),
+    ],
+)
+def test_logs_json_body_function_argument_shape_errors(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    expression: str,
+    expected_error: str,
+) -> None:
+    """has-family value arguments are shape-validated; mis-shaped arguments are rejected (400)."""
+    now = datetime.now(tz=UTC)
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    response = make_query_request(
+        signoz,
+        token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+    )
+    assert response.status_code == HTTPStatus.BAD_REQUEST, f"{expression}: expected 400, got {response.status_code}: {response.text}"
+    assert expected_error in response.text, f"{expression}: expected error {expected_error!r} in {response.text}"
+
+
+def test_logs_json_body_type_match_by_needle(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """The element type is inferred from the needle (legacy has no schema): a string needle
+    searches as String, a quoted-integer needle as Int64, a decimal needle as Float64. A
+    numeric-string body array (["100","200"]) is parsed, so numeric needles still match it."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=4), resources={"service.name": "app-service"}, body=json.dumps({"vals": ["prod", "api"]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body=json.dumps({"vals": [100, 200, 300]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"vals": [1.5, 2.5]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"vals": ["100", "200"]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [
+        ("has(body.vals, 'prod')", 1),  # String -> only the string array
+        ('has(body.vals, "200")', 2),  # quoted Int64 -> [100,200,300] and ["100","200"]
+        ("has(body.vals, 1.5)", 1),  # Float64 -> only the float array
+    ]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"
+
+
+def test_logs_json_body_large_integer_exact(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """Large ids (> 2^53) stay exact only when QUOTED (extracted as Int64). An unquoted numeric
+    literal is parsed as float64 upstream, so it false-matches neighbouring ids (documented)."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"id": [1234567890123456789]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"id": [1234567890123456788]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    # Quoted -> Int64 -> exact: matches only the exact id, not the neighbour.
+    response = make_query_request(
+        signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='has(body.id, "1234567890123456789")')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 1, "quoted big id should match exactly one log"
+    assert json.loads(rows[0]["data"]["body"])["id"] == [1234567890123456789]
+
+    # Unquoted -> float64 upstream -> imprecise: false-matches the off-by-one neighbour too.
+    response = make_query_request(
+        signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.id, 1234567890123456789)")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    assert len(get_rows(response)) == 2, "unquoted big id collides with neighbour (float64 precision) - quote large ids"
+
+
+def test_logs_json_body_no_supertype_safety(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """A needle whose type doesn't match the body array must never raise a ClickHouse "no
+    supertype" error (code 386): it returns 200 with no match (Nullable extraction -> NULL)."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"names": ["alice", "bob"]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"nums": [100, 200]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression in ["has(body.names, 200)", "has(body.nums, 'alice')"]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: expected 200 (no supertype error), got {response.status_code}: {response.text}"
+        assert get_rows(response) == [], f"{expression}: type mismatch should match nothing"
+
+
+def test_logs_json_body_mixed_needle_args(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """A value list mixing types (e.g. [200, "abc"]) is NOT rejected; it falls back to a string
+    comparison (safe, no error). Numbers render to their text form, so both logs match."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"vals": [100, 200, 300]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"vals": ["abc", "xyz"]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    response = make_query_request(
+        signoz, token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression='hasAny(body.vals, [200, "abc"])')],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    assert len(get_rows(response)) == 2, "mixed-type args fall back to string comparison; both logs match"
+
+
+def test_logs_json_body_nested_array_path(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """has over a path that traverses arrays (body.edu[*].names) searches the flattened leaf."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"edu": [{"names": ["IIT", "MIT"]}, {"names": ["CMU"]}]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"edu": [{"names": ["Stanford"]}]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [("has(body.edu[*].names, 'MIT')", 1), ("has(body.edu[*].names, 'Yale')", 0)]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"
+
+
+def test_logs_json_body_missing_empty_null(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """has matches nothing for a missing key, an empty array, or a null value (only the log
+    that actually contains the value matches)."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=4), resources={"service.name": "app-service"}, body=json.dumps({"other": [1, 2]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body=json.dumps({"vals": []}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"vals": None}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"vals": ["x"]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    response = make_query_request(
+        signoz, token,
+        start_ms=int((now - timedelta(seconds=10)).timestamp() * 1000),
+        end_ms=int(now.timestamp() * 1000),
+        request_type=RequestType.RAW,
+        queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression="has(body.vals, 'x')")],
+    )
+    assert response.status_code == HTTPStatus.OK, response.text
+    rows = get_rows(response)
+    assert len(rows) == 1, "only the log whose vals contains 'x' matches"
+    assert json.loads(rows[0]["data"]["body"])["vals"] == ["x"]
+
+
+def test_logs_json_body_negative_and_special_chars(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """Negative integers and strings with spaces/special chars match correctly."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"nums": [-5, -10]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"tags": ["a b", "c/d"]}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [('has(body.nums, "-5")', 1), ("has(body.tags, 'a b')", 1)]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"
+
+
+def test_logs_json_body_type_isolation(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_logs: Callable[[list[Logs]], None],
+) -> None:
+    """Type isolation (regression guard). A bool needle matches only genuine booleans, not
+    "truthy" numbers/strings; an empty-string or zero needle does not false-match array bodies;
+    genuine scalar bool / zero / empty-string bodies still match (the single-element-set path)."""
+    now = datetime.now(tz=UTC)
+    insert_logs(
+        [
+            Logs(timestamp=now - timedelta(seconds=5), resources={"service.name": "app-service"}, body=json.dumps({"nums": [100, 200, 300]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=4), resources={"service.name": "app-service"}, body=json.dumps({"flags": [True, False]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=3), resources={"service.name": "app-service"}, body=json.dumps({"tags": ["a", "b"]}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=2), resources={"service.name": "app-service"}, body=json.dumps({"flag": True}), severity_text="INFO"),
+            Logs(timestamp=now - timedelta(seconds=1), resources={"service.name": "app-service"}, body=json.dumps({"code": 0}), severity_text="INFO"),
+        ]
+    )
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    start_ms = int((now - timedelta(seconds=10)).timestamp() * 1000)
+    end_ms = int(now.timestamp() * 1000)
+
+    for expression, expected in [
+        ("has(body.nums, true)", 0),  # bool needle does NOT truthy-match a numeric array
+        ("has(body.flags, true)", 1),  # genuine bool array matches
+        ("has(body.flag, true)", 1),  # genuine scalar bool matches
+        ("has(body.tags, '')", 0),  # empty-string does NOT false-match an array body
+        ("has(body.nums, 0)", 0),  # zero does NOT false-match an array body without a 0
+        ("has(body.code, 0)", 1),  # genuine scalar zero matches
+    ]:
+        response = make_query_request(
+            signoz, token, start_ms=start_ms, end_ms=end_ms, request_type=RequestType.RAW,
+            queries=[build_raw_query("A", "logs", order=[build_order_by("timestamp")], limit=100, filter_expression=expression)],
+        )
+        assert response.status_code == HTTPStatus.OK, f"{expression}: {response.text}"
+        assert len(get_rows(response)) == expected, f"{expression}: expected {expected} rows"