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 structured extraction requests function_calling

dex0shubhamaa047083f761
diff --git a/tests/types/test_structured_output_method.py b/tests/types/test_structured_output_method.py
new file mode 100644
index 0000000..dfdfcd0
--- /dev/null
+++ b/tests/types/test_structured_output_method.py
@@ -0,0 +1,58 @@
+"""Structured extraction must request method="function_calling" (issue #52).
+
+langchain-openai defaults to method="json_schema", which only OpenAI supports;
+DeepSeek / vLLM / Ollama and other OpenAI-compatible endpoints reject it.
+"""
+
+from pydantic import BaseModel
+
+from tests.mocks import MockEmbeddings, MockStructuredRunnable
+from hyperextract.types import AutoModel, AutoGraph
+
+
+class _Person(BaseModel):
+ name: str
+
+
+class _Entity(BaseModel):
+ name: str
+ type: str = "x"
+
+
+class _Relation(BaseModel):
+ source: str
+ target: str
+
+
+class _SpyLLM:
+ """Records the `method` passed to each with_structured_output call."""
+
+ def __init__(self):
+ self.captured = []
+
+ def with_structured_output(self, schema, **kwargs):
+ self.captured.append(kwargs.get("method"))
+ return MockStructuredRunnable(schema=schema)
+
+
+def test_automodel_requests_function_calling():
+ spy = _SpyLLM()
+ AutoModel(data_schema=_Person, llm_client=spy, embedder=MockEmbeddings())
+ assert spy.captured
+ assert all(m == "function_calling" for m in spy.captured)
+
+
+def test_autograph_two_stage_requests_function_calling():
+ spy = _SpyLLM()
+ AutoGraph(
+ node_schema=_Entity,
+ edge_schema=_Relation,
+ node_key_extractor=lambda x: x.name,
+ edge_key_extractor=lambda x: f"{x.source}-{x.target}",
+ nodes_in_edge_extractor=lambda x: (x.source, x.target),
+ llm_client=spy,
+ embedder=MockEmbeddings(),
+ extraction_mode="two_stage",
+ )
+ assert spy.captured
+ assert all(m == "function_calling" for m in spy.captured)