claude-sonnet-5 review Phase 3 (by gpt-5.3-codex)

tests/services/test_store.py

@@ -10,9 +10,11 @@ from transcription.db.models import Document
 from transcription.db.models import Job
 from transcription.db.models import JobSource
 from transcription.db.models import Source
+from transcription.errors import ErrorCategory
 from transcription.services.people import store_person_portrait
 from transcription.services.sources import source_mime_type
 from transcription.services.store import SourceStorageError
+from transcription.services.store import StoredSourceFile
 from transcription.services.store import create_document_job
 from transcription.services.store import create_job_for_document
 from transcription.services.store import store_source_file
@@ -152,3 +154,72 @@ async def test_source_storage_rejects_unsupported_format(tmp_path):
 
     with pytest.raises(SourceStorageError):
         await store_source_file(filename="page.txt", file_bytes=b"text", settings=settings)
+
+
+@pytest.mark.asyncio
+async def test_create_document_job_db_failure_maps_to_internal_category(tmp_path, monkeypatch):
+    settings = Settings(openrouter_api_key="test-key", upload_dir=tmp_path)
+    stored_path = tmp_path / "documents" / "stored-file.pdf"
+    stored_path.parent.mkdir(parents=True, exist_ok=True)
+    stored_path.write_bytes(b"payload")
+
+    async def _fake_store_source_file(**_kwargs) -> StoredSourceFile:
+        return StoredSourceFile(
+            path=stored_path,
+            file_hash="f" * 64,
+            file_size_bytes=7,
+        )
+
+    async def _boom_create_records(**_kwargs):
+        raise RuntimeError("database unavailable")
+
+    monkeypatch.setattr("transcription.services.store.store_source_file", _fake_store_source_file)
+    monkeypatch.setattr("transcription.services.store._create_document_job_records", _boom_create_records)
+
+    with pytest.raises(SourceStorageError) as exc_info:
+        await create_document_job(
+            filename="single-page.pdf",
+            file_bytes=b"payload",
+            settings=settings,
+        )
+
+    assert exc_info.value.category == ErrorCategory.INFRA_PERSISTENT
+    assert exc_info.value.retriable is False
+    assert not stored_path.exists()
+
+
+@pytest.mark.asyncio
+async def test_create_job_for_document_db_failure_maps_to_internal_category(async_session, tmp_path, monkeypatch):
+    document = Document(id=uuid4(), name="db-failure-doc")
+    async_session.add(document)
+    await async_session.commit()
+
+    settings = Settings(openrouter_api_key="test-key", upload_dir=tmp_path)
+    stored_path = tmp_path / "documents" / str(document.id) / "stored-file.pdf"
+    stored_path.parent.mkdir(parents=True, exist_ok=True)
+    stored_path.write_bytes(b"payload")
+
+    async def _fake_store_source_file(**_kwargs) -> StoredSourceFile:
+        return StoredSourceFile(
+            path=stored_path,
+            file_hash="f" * 64,
+            file_size_bytes=7,
+        )
+
+    async def _boom_create_records(**_kwargs):
+        raise RuntimeError("database unavailable")
+
+    monkeypatch.setattr("transcription.services.store.store_source_file", _fake_store_source_file)
+    monkeypatch.setattr("transcription.services.store._create_job_for_document_records", _boom_create_records)
+
+    with pytest.raises(SourceStorageError) as exc_info:
+        await create_job_for_document(
+            document_id=document.id,
+            source_files=[("single-page.pdf", b"payload")],
+            session=async_session,
+            settings=settings,
+        )
+
+    assert exc_info.value.category == ErrorCategory.INFRA_PERSISTENT
+    assert exc_info.value.retriable is False
+    assert not stored_path.exists()