Skip to content

Hypergraph is more powerful. Transform unstructured text into structured knowledge with LLMs. Graphs, hypergraphs, and spatio-temporal extractions — with one command.

test: assert load_metadata restores datetime timestamps

dex0shubham1c2ca0f82dda
diff --git a/tests/types/test_model_serialization.py b/tests/types/test_model_serialization.py
index a06cc5c..1c0d9a6 100644
--- a/tests/types/test_model_serialization.py
+++ b/tests/types/test_model_serialization.py
@@ -140,6 +140,36 @@ class TestAutoModelSerialization:
assert "created_at" in new_model.metadata
assert "updated_at" in new_model.metadata
+ def test_load_metadata_restores_datetime_type(self, llm_client, embedder):
+ """Loaded timestamps are datetime objects, not ISO strings.
+
+ dump_metadata serializes datetimes via default=str; without conversion
+ on load, a loaded instance keeps strings while a fresh one holds
+ datetimes, so min(created_at, ...) in __add__ would raise TypeError.
+ """
+ from datetime import datetime
+
+ model = AutoModel(
+ data_schema=PersonSchema,
+ llm_client=llm_client,
+ embedder=embedder,
+ )
+
+ metadata_file = Path(self.temp_dir) / "metadata.json"
+ model.dump_metadata(metadata_file)
+
+ new_model = AutoModel(
+ data_schema=PersonSchema,
+ llm_client=llm_client,
+ embedder=embedder,
+ )
+ new_model.load_metadata(metadata_file)
+
+ assert isinstance(new_model.metadata["created_at"], datetime)
+ assert isinstance(new_model.metadata["updated_at"], datetime)
+ # The real-world manifestation: comparing loaded vs fresh must not raise.
+ assert min(new_model.metadata["created_at"], datetime.now())
+
@pytest.mark.skipif(not os.getenv("OPENAI_API_KEY"), reason="Requires real LLM")
def test_dump_and_load_full(self, llm_client, embedder):
"""Test full dump/load cycle."""