Require concrete schema strings in extract tool params

Co-authored-by: Shri Sukhani <shrisukhani@users.noreply.github.com>

Author: cursoragent

hyperbrowser/tools/__init__.py

@@ -104,11 +104,13 @@ def _normalize_extract_schema_mapping(
 def _prepare_extract_tool_params(params: Mapping[str, Any]) -> Dict[str, Any]:
     normalized_params = _to_param_dict(params)
     schema_value = normalized_params.get("schema")
-    if schema_value is not None and not isinstance(schema_value, (str, MappingABC)):
+    if schema_value is not None and not (
+        type(schema_value) is str or isinstance(schema_value, MappingABC)
+    ):
         raise HyperbrowserError(
             "Extract tool `schema` must be an object or JSON string"
         )
-    if isinstance(schema_value, str):
+    if type(schema_value) is str:
         try:
             parsed_schema = json.loads(schema_value)
         except HyperbrowserError:

tests/test_tools_extract.py

@@ -174,6 +174,43 @@ async def run():
         asyncio.run(run())
 
 
+def test_extract_tool_runnable_rejects_string_subclass_schema_strings():
+    class _SchemaString(str):
+        pass
+
+    client = _SyncClient()
+    params = {
+        "urls": ["https://example.com"],
+        "schema": _SchemaString('{"type":"object"}'),
+    }
+
+    with pytest.raises(
+        HyperbrowserError,
+        match="Extract tool `schema` must be an object or JSON string",
+    ):
+        WebsiteExtractTool.runnable(client, params)
+
+
+def test_extract_tool_async_runnable_rejects_string_subclass_schema_strings():
+    class _SchemaString(str):
+        pass
+
+    client = _AsyncClient()
+    params = {
+        "urls": ["https://example.com"],
+        "schema": _SchemaString('{"type":"object"}'),
+    }
+
+    async def run():
+        await WebsiteExtractTool.async_runnable(client, params)
+
+    with pytest.raises(
+        HyperbrowserError,
+        match="Extract tool `schema` must be an object or JSON string",
+    ):
+        asyncio.run(run())
+
+
 def test_extract_tool_runnable_rejects_null_schema_json():
     client = _SyncClient()
     params = {