refactor(tests): Update tests to remove summary references and adjust for embedding errors

surfsense_backend/tests/integration/conftest.py

@@ -1,7 +1,7 @@
 import importlib
 import sys
 import uuid
-from unittest.mock import AsyncMock, MagicMock
+from unittest.mock import MagicMock
 
 import pytest
 import pytest_asyncio
@@ -123,18 +123,6 @@ async def db_search_space(db_session: AsyncSession, db_user: User) -> SearchSpac
     return space
 
 
-@pytest.fixture
-def patched_summarize(monkeypatch) -> AsyncMock:
-    mock = AsyncMock(return_value="Mocked summary.")
-    return mock
-
-
-@pytest.fixture
-def patched_summarize_raises(monkeypatch) -> AsyncMock:
-    mock = AsyncMock(side_effect=RuntimeError("LLM unavailable"))
-    return mock
-
-
 @pytest.fixture
 def patched_embed_texts(monkeypatch) -> MagicMock:
     mock = MagicMock(side_effect=lambda texts: [[0.1] * _EMBEDDING_DIM for _ in texts])
@@ -145,6 +133,16 @@ def patched_embed_texts(monkeypatch) -> MagicMock:
     return mock
 
 
+@pytest.fixture
+def patched_embed_texts_raises(monkeypatch) -> MagicMock:
+    mock = MagicMock(side_effect=RuntimeError("Embedding unavailable"))
+    monkeypatch.setattr(
+        "app.indexing_pipeline.indexing_pipeline_service.embed_texts",
+        mock,
+    )
+    return mock
+
+
 @pytest.fixture
 def patched_chunk_text(monkeypatch) -> MagicMock:
     mock = MagicMock(return_value=["Test chunk content."])

surfsense_backend/tests/integration/indexing_pipeline/adapters/test_file_upload_adapter.py

@@ -32,8 +32,8 @@ async def test_sets_status_ready(db_session, db_search_space, db_user, mocker):
 @pytest.mark.usefixtures(
 "patched_embed_texts", "patched_chunk_text"
 )
-async def test_content_is_summary(db_session, db_search_space, db_user, mocker):
-    """Document content is set to the LLM-generated summary."""
+async def test_content_is_source_markdown(db_session, db_search_space, db_user, mocker):
+    """Document content is set to the extracted source markdown."""
     adapter = UploadDocumentAdapter(db_session)
     await adapter.index(
         markdown_content="## Hello\n\nSome content.",
@@ -48,7 +48,7 @@ async def test_content_is_summary(db_session, db_search_space, db_user, mocker):
     )
     document = result.scalars().first()
 
-    assert document.content == "Mocked summary."
+    assert document.content == "## Hello\n\nSome content."
 
 
 @pytest.mark.usefixtures(
@@ -79,9 +79,7 @@ async def test_chunks_written_to_db(db_session, db_search_space, db_user, mocker
     assert chunks[0].content == "Test chunk content."
 
 
-@pytest.mark.usefixtures(
-    "patched_summarize_raises", "patched_embed_texts", "patched_chunk_text"
-)
+@pytest.mark.usefixtures("patched_embed_texts_raises", "patched_chunk_text")
 async def test_raises_on_indexing_failure(db_session, db_search_space, db_user, mocker):
     """RuntimeError is raised when the indexing step fails so the caller can fire a failure notification."""
     adapter = UploadDocumentAdapter(db_session)
@@ -92,7 +90,7 @@ async def test_raises_on_indexing_failure(db_session, db_search_space, db_user,
             etl_service="UNSTRUCTURED",
             search_space_id=db_search_space.id,
             user_id=str(db_user.id),
-                )
+        )
 
 
 # ---------------------------------------------------------------------------
@@ -104,7 +102,7 @@ async def test_raises_on_indexing_failure(db_session, db_search_space, db_user,
 "patched_embed_texts", "patched_chunk_text"
 )
 async def test_reindex_updates_content(db_session, db_search_space, db_user, mocker):
-    """Document content is updated to the new summary after reindexing."""
+    """Document content is updated to the new source markdown after reindexing."""
     adapter = UploadDocumentAdapter(db_session)
     await adapter.index(
         markdown_content="## Original\n\nOriginal content.",
@@ -125,7 +123,7 @@ async def test_reindex_updates_content(db_session, db_search_space, db_user, moc
     await adapter.reindex(document=document)
 
     await db_session.refresh(document)
-    assert document.content == "Mocked summary."
+    assert document.content == "## Edited\n\nNew content after user edit."
 
 
 @pytest.mark.usefixtures(
@@ -256,7 +254,9 @@ async def test_reindex_clears_reindexing_flag(
 
 
 @pytest.mark.usefixtures("patched_embed_texts", "patched_chunk_text")
-async def test_reindex_raises_on_failure(db_session, db_search_space, db_user, mocker):
+async def test_reindex_raises_on_failure(
+    db_session, db_search_space, db_user, patched_embed_texts, mocker
+):
     """RuntimeError is raised when reindexing fails so the caller can handle it."""
 
     adapter = UploadDocumentAdapter(db_session)
@@ -276,6 +276,8 @@ async def test_reindex_raises_on_failure(db_session, db_search_space, db_user, m
     document.source_markdown = "## Edited\n\nNew content after user edit."
     await db_session.flush()
 
+    patched_embed_texts.side_effect = RuntimeError("Embedding unavailable")
+
     with pytest.raises(RuntimeError, match=r"Embedding failed|Reindexing failed"):
         await adapter.reindex(document=document)
 

surfsense_backend/tests/integration/indexing_pipeline/test_index_document.py

@@ -259,16 +259,14 @@ async def test_reindex_replaces_old_chunks(
     assert len(chunks) == 1
 
 
-@pytest.mark.usefixtures(
-    "patched_summarize_raises", "patched_embed_texts", "patched_chunk_text"
-)
-async def test_llm_error_sets_status_failed(
+@pytest.mark.usefixtures("patched_embed_texts_raises", "patched_chunk_text")
+async def test_embedding_error_sets_status_failed(
     db_session,
     db_search_space,
     make_connector_document,
     mocker,
 ):
-    """Document status is FAILED when the LLM raises during indexing."""
+    """Document status is FAILED when embedding raises during indexing."""
     connector_doc = make_connector_document(search_space_id=db_search_space.id)
     service = IndexingPipelineService(session=db_session)
 
@@ -286,10 +284,8 @@ async def test_llm_error_sets_status_failed(
     assert DocumentStatus.is_state(reloaded.status, DocumentStatus.FAILED)
 
 
-@pytest.mark.usefixtures(
-    "patched_summarize_raises", "patched_embed_texts", "patched_chunk_text"
-)
-async def test_llm_error_leaves_no_partial_data(
+@pytest.mark.usefixtures("patched_embed_texts_raises", "patched_chunk_text")
+async def test_embedding_error_leaves_no_partial_data(
     db_session,
     db_search_space,
     make_connector_document,

surfsense_backend/tests/integration/indexing_pipeline/test_local_folder_pipeline.py

@@ -1271,7 +1271,7 @@ class TestIndexingProgressFlag:
         original_index = IndexingPipelineService.index
         flag_observed = []
 
-        async def patched_index(self_pipe, document, connector_doc, llm):
+        async def patched_index(self_pipe, document, connector_doc):
             folder = (
                 await db_session.execute(
                     select(Folder).where(
@@ -1283,7 +1283,7 @@ class TestIndexingProgressFlag:
             if folder:
                 meta = folder.folder_metadata or {}
                 flag_observed.append(meta.get("indexing_in_progress", False))
-            return await original_index(self_pipe, document, connector_doc, llm)
+            return await original_index(self_pipe, document, connector_doc)
 
         IndexingPipelineService.index = patched_index
         try:

surfsense_backend/tests/integration/indexing_pipeline/test_prepare_for_indexing.py

@@ -338,9 +338,7 @@ async def test_same_content_from_different_source_is_skipped(
     assert len(result.scalars().all()) == 1
 
 
-@pytest.mark.usefixtures(
-    "patched_summarize_raises", "patched_embed_texts", "patched_chunk_text"
-)
+@pytest.mark.usefixtures("patched_embed_texts_raises", "patched_chunk_text")
 async def test_failed_document_with_unchanged_content_is_requeued(
     db_session,
     db_search_space,
@@ -351,7 +349,7 @@ async def test_failed_document_with_unchanged_content_is_requeued(
     doc = make_connector_document(search_space_id=db_search_space.id)
     service = IndexingPipelineService(session=db_session)
 
-    # First run: document is created and indexing crashes → status = failed
+    # First run: document is created and indexing crashes, so status becomes failed.
     prepared = await service.prepare_for_indexing([doc])
     document_id = prepared[0].id
     await service.index(prepared[0], doc)