fix: validate top-level JSON arrays in JsonSchemaValidator

haystack/components/validators/json_schema.py

@@ -161,12 +161,18 @@ def run(
         else:
             validation_schema = json_schema
         try:
-            last_message_json = [last_message_json] if not isinstance(last_message_json, list) else last_message_json
-            for content in last_message_json:
-                if using_openai_schema:
-                    validate(instance=content["function"]["arguments"], schema=validation_schema)
-                else:
-                    validate(instance=content, schema=validation_schema)
+            is_function_calling_payload = using_openai_schema or self._is_function_calling_message_list(
+                last_message_json
+            )
+            if is_function_calling_payload:
+                last_message_json = last_message_json if isinstance(last_message_json, list) else [last_message_json]
+                for content in last_message_json:
+                    if using_openai_schema:
+                        validate(instance=content["function"]["arguments"], schema=validation_schema)
+                    else:
+                        validate(instance=content, schema=validation_schema)
+            else:
+                validate(instance=last_message_json, schema=validation_schema)
 
             return {"validated": [last_message]}
         except ValidationError as e:
@@ -218,6 +224,24 @@ def _is_openai_function_calling_schema(self, json_schema: dict[str, Any]) -> boo
         """
         return all(key in json_schema for key in ["name", "description", "parameters"])
 
+    def _is_function_calling_message_list(self, data: Any) -> bool:
+        """
+        Checks if the provided data is a non-empty list of OpenAI function calling messages.
+
+        :param data: The data to check.
+        :return: `True` if the data is a function calling message list; otherwise, `False`.
+        """
+        if not isinstance(data, list) or not data:
+            return False
+
+        for message in data:
+            if not isinstance(message, dict):
+                return False
+            function = message.get("function")
+            if not isinstance(function, dict) or "arguments" not in function:
+                return False
+        return True
+
     def _recursive_json_to_object(self, data: Any) -> Any:
         """
         Convert any string values that are valid JSON objects into dictionary objects.

releasenotes/notes/fix-json-schema-validator-array-validation-05344b7b7b18680b.yaml

@@ -0,0 +1,7 @@
+---
+fixes:
+  - |
+    Fixed ``JsonSchemaValidator`` rejecting valid top-level JSON arrays when the provided schema describes an array.
+    Regular schemas now validate the complete payload once, while OpenAI function-calling payloads continue to
+    validate each tool call individually. Array constraints such as ``minItems`` are now also enforced for empty
+    arrays.

test/components/validators/test_json_schema.py

@@ -141,6 +141,24 @@ def test_validation_error_for_top_level_json_scalar(self, message_text):
         assert "validation_error" in result
         assert len(result["validation_error"]) == 1
 
+    def test_validates_top_level_json_array(self):
+        validator = JsonSchemaValidator(json_schema={"type": "array", "items": {"type": "integer"}})
+        message = ChatMessage.from_assistant("[1, 2]")
+
+        result = validator.run([message])
+
+        assert result == {"validated": [message]}
+
+    def test_rejects_empty_top_level_json_array(self):
+        validator = JsonSchemaValidator(json_schema={"type": "array", "items": {"type": "integer"}, "minItems": 1})
+        message = ChatMessage.from_assistant("[]")
+
+        result = validator.run([message])
+
+        assert "validation_error" in result
+        assert len(result["validation_error"]) == 1
+        assert "should be non-empty" in result["validation_error"][0].text
+
     #  Validates multiple messages against a provided JSON schema successfully.
     def test_validates_multiple_messages_against_json_schema(self, json_schema_github_compare, genuine_fc_message):
         validator = JsonSchemaValidator()