fix: only set additionalProperties on object schemas

pyproject.toml

@@ -33,7 +33,6 @@ dependencies = [
    "typing_extensions",
    "outlines_core==0.2.14",
    "genson",
-   "jsonpath_ng",
 ]
 dynamic = ["version"]
 
@@ -180,7 +179,6 @@ module = [
     "tf-keras",
     "tf-keras.*",
     "mkdocs_gen_files.*",
-    "jsonpath_ng.*",
     "llguidance.*",
     "xgrammar.*",
 ]

src/outlines/models/utils.py

@@ -1,8 +1,10 @@
-import jsonpath_ng
-
-
 def set_additional_properties_false_json_schema(schema: dict) -> dict:
-    """Set additionalProperties to False to all objects in the schema using jsonpath.
+    """Set additionalProperties to False on all object schemas.
+
+    Recursively walks the JSON schema and, for every object schema, sets
+    ``additionalProperties`` to False unless it is already present. An object
+    schema is one whose ``type`` is ``"object"`` or, for a nullable object, a
+    list containing ``"object"`` (e.g. ``["object", "null"]``).
 
     Parameters
     ----------
@@ -14,20 +16,26 @@ def set_additional_properties_false_json_schema(schema: dict) -> dict:
     dict
         The modified schema with additionalProperties set to False
     """
-    # Get all nodes
-    jsonpath_expr = jsonpath_ng.parse('$..*')
-    matches = jsonpath_expr.find(schema)
 
-    # Go over all nodes and set additionalProperties to False if it's an
-    # object. `type` can either be the bare string "object" or, per the JSON
-    # Schema spec, a list of type names (e.g. ["object", "null"]) used to
-    # express nullable objects, so both forms must be checked.
-    for match in matches:
-        is_object_type = match.value == 'object' or (
-            isinstance(match.value, list) and 'object' in match.value
-        )
-        if is_object_type:
-            if 'additionalProperties' not in match.context.value:
-                match.context.value['additionalProperties'] = False
+    def _walk(node):
+        if isinstance(node, dict):
+            # Only an object *schema* should get ``additionalProperties``. Keying
+            # off the string ``"object"`` appearing anywhere would wrongly add the
+            # keyword to non-object schemas, e.g. a string field with
+            # ``const``/``default``/``title`` equal to ``"object"``, which the
+            # OpenAI API then rejects. ``type`` may be the bare string or a list
+            # such as ``["object", "null"]`` for a nullable object.
+            node_type = node.get("type")
+            is_object = node_type == "object" or (
+                isinstance(node_type, list) and "object" in node_type
+            )
+            if is_object and "additionalProperties" not in node:
+                node["additionalProperties"] = False
+            for value in node.values():
+                _walk(value)
+        elif isinstance(node, list):
+            for item in node:
+                _walk(item)
 
+    _walk(schema)
     return schema

tests/models/test_utils.py

@@ -51,7 +51,7 @@ def test_set_additional_properties_false_json_schema():
 
 
 def test_set_additional_properties_false_json_schema_nested_object():
-    # nested object with a plain "type": "object" gets additionalProperties too
+    # nested object schemas get additionalProperties too
     schema = {
         "type": "object",
         "properties": {
@@ -88,3 +88,27 @@ def test_set_additional_properties_false_json_schema_nullable_type_array():
     modified_schema = set_additional_properties_false_json_schema(schema)
     assert modified_schema["additionalProperties"] is False
     assert modified_schema["properties"]["address"]["additionalProperties"] is False
+
+
+def test_set_additional_properties_false_json_schema_string_value_object():
+    # A non-object schema whose value happens to equal the string "object"
+    # (e.g. a Literal["object"] field emitted by pydantic as `const`) must NOT
+    # get `additionalProperties`, which the OpenAI API rejects on non-objects.
+    schema = {
+        "type": "object",
+        "properties": {
+            "kind": {"type": "string", "const": "object", "title": "object"},
+        },
+        "required": ["kind"],
+    }
+    modified_schema = set_additional_properties_false_json_schema(schema)
+    target_schema = {
+        "type": "object",
+        "properties": {
+            "kind": {"type": "string", "const": "object", "title": "object"},
+        },
+        "required": ["kind"],
+        "additionalProperties": False,
+    }
+    assert modified_schema == target_schema
+    assert "additionalProperties" not in modified_schema["properties"]["kind"]