Fixed S2 crash on max over empty sequence

Author: jamesbraza

src/paperqa/clients/semantic_scholar.py

@@ -265,7 +265,9 @@ async def s2_title_search(
             (strings_similarity(entry["title"], title), entry)
             for entry in data.get("data", data)
         )
-    except (KeyError, IndexError) as exc:
+    except (KeyError, IndexError, ValueError) as exc:
+        # ValueError: S2 may return {"data": []} causing max() on an empty iterable to
+        # throw a ValueError
         raise DOINotFoundError(
             f"Unexpected Semantic Scholar search/match endpoint shape for {title}"
             f" given data {data}."

tests/cassettes/test_s2_title_search_empty_data.yaml

@@ -22,12 +22,14 @@
     SemanticScholarProvider,
 )
 from paperqa.clients.client_models import MetadataPostProcessor, MetadataProvider
+from paperqa.clients.exceptions import DOINotFoundError
 from paperqa.clients.journal_quality import (
     DEFAULT_JOURNAL_QUALITY_CSV_PATH,
     JournalQualityPostProcessor,
 )
 from paperqa.clients.openalex import OpenAlexProvider, reformat_name
 from paperqa.clients.retractions import RetractionDataPostProcessor
+from paperqa.clients.semantic_scholar import s2_title_search
 from paperqa.types import SOURCE_QUALITY_MESSAGES, DocDetails
 
 # Use to avoid flaky tests every time citation count changes
@@ -379,6 +381,32 @@ async def test_client_os_error() -> None:
         assert mock_get.call_count >= 1, "Expected the exception to have been thrown"
 
 
+@pytest.mark.asyncio
+@pytest.mark.parametrize(
+    "mock_return_data",
+    [
+        pytest.param({"data": []}, id="empty-data-list"),
+    ],
+)
+async def test_s2_title_search_edge_cases(mock_return_data: dict) -> None:
+    async with httpx_aiohttp.HttpxAiohttpClient() as http_client:
+        with patch(
+            "paperqa.clients.semantic_scholar._s2_get_with_retrying",
+            return_value=mock_return_data,
+        ):
+            with pytest.raises(DOINotFoundError):
+                await s2_title_search("some title", client=http_client)
+
+
+@pytest.mark.vcr
+@pytest.mark.asyncio
+async def test_s2_title_search_empty_data() -> None:
+    """Confirm an S2 match response with empty data raises DOINotFoundError."""
+    async with httpx_aiohttp.HttpxAiohttpClient() as http_client:
+        with pytest.raises(DOINotFoundError):
+            await s2_title_search("empty results edge case query", client=http_client)
+
+
 @pytest.mark.vcr
 @pytest.mark.asyncio
 async def test_bad_dois() -> None: