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: parse() survives extractor failures

dex0shubham68216b992b99
diff --git a/tests/types/test_forgiving_extraction.py b/tests/types/test_forgiving_extraction.py
new file mode 100644
index 0000000..0fe1bf0
--- /dev/null
+++ b/tests/types/test_forgiving_extraction.py
@@ -0,0 +1,43 @@
+"""Extraction tolerates LLM parse failures instead of crashing (issue #45)."""
+
+from pydantic import BaseModel
+
+from tests.mocks import MockChatModel, MockEmbeddings
+from hyperextract.types import AutoModel
+
+
+class _Person(BaseModel):
+ name: str = ""
+
+
+class _BoomExtractor:
+ """Stands in for a data_extractor whose LLM returns unparseable output."""
+
+ def invoke(self, _input, config=None):
+ raise ValueError("Invalid JSON: expected value at line 1 column 1")
+
+ def batch(self, inputs, config=None, return_exceptions=False, **kwargs):
+ errs = [ValueError("Invalid JSON") for _ in inputs]
+ if return_exceptions:
+ return errs
+ raise errs[0]
+
+
+def _model():
+ m = AutoModel(
+ data_schema=_Person, llm_client=MockChatModel(), embedder=MockEmbeddings()
+ )
+ m.data_extractor = _BoomExtractor()
+ return m
+
+
+def test_parse_survives_single_chunk_failure():
+ # text <= chunk_size -> invoke() path; a parse error yields an empty result
+ result = _model().parse("short text")
+ assert result.empty()
+
+
+def test_parse_survives_multi_chunk_failure():
+ # text > chunk_size -> batch() path
+ result = _model().parse("sentence. " * 400)
+ assert result.empty()