fix: correct test fixtures and add missing summarizer tests

surfsense_backend/tests/unit/indexing_pipeline/conftest.py

@@ -1,15 +1,33 @@
 import pytest
+from unittest.mock import AsyncMock, MagicMock
 
 
 @pytest.fixture
-def patched_chunker_instance(mocker):
-    mock = mocker.patch("app.indexing_pipeline.document_chunker.config.chunker_instance")
-    mock.chunk.return_value = [mocker.Mock(text="prose chunk")]
+def patched_summarizer_chain(monkeypatch):
+    chain = MagicMock()
+    chain.ainvoke = AsyncMock(return_value=MagicMock(content="The summary."))
+
+    template = MagicMock()
+    template.__or__ = MagicMock(return_value=chain)
+
+    monkeypatch.setattr(
+        "app.indexing_pipeline.document_summarizer.SUMMARY_PROMPT_TEMPLATE",
+        template,
+    )
+    return chain
+
+
+@pytest.fixture
+def patched_chunker_instance(monkeypatch):
+    mock = MagicMock()
+    mock.chunk.return_value = [MagicMock(text="prose chunk")]
+    monkeypatch.setattr("app.indexing_pipeline.document_chunker.config.chunker_instance", mock)
     return mock
 
 
 @pytest.fixture
-def patched_code_chunker_instance(mocker):
-    mock = mocker.patch("app.indexing_pipeline.document_chunker.config.code_chunker_instance")
-    mock.chunk.return_value = [mocker.Mock(text="code chunk")]
+def patched_code_chunker_instance(monkeypatch):
+    mock = MagicMock()
+    mock.chunk.return_value = [MagicMock(text="code chunk")]
+    monkeypatch.setattr("app.indexing_pipeline.document_chunker.config.code_chunker_instance", mock)
     return mock

surfsense_backend/tests/unit/indexing_pipeline/test_document_chunker.py

@@ -5,15 +5,15 @@ from app.indexing_pipeline.document_chunker import chunk_text
 pytestmark = pytest.mark.unit
 
 
-def test_uses_code_chunker_when_flag_is_true(patched_code_chunker_instance):
+@pytest.mark.usefixtures("patched_chunker_instance", "patched_code_chunker_instance")
+def test_uses_code_chunker_when_flag_is_true():
     result = chunk_text("def foo(): pass", use_code_chunker=True)
 
-    patched_code_chunker_instance.chunk.assert_called_once_with("def foo(): pass")
     assert result == ["code chunk"]
 
 
-def test_uses_default_chunker_when_flag_is_false(patched_chunker_instance):
+@pytest.mark.usefixtures("patched_chunker_instance", "patched_code_chunker_instance")
+def test_uses_default_chunker_when_flag_is_false():
     result = chunk_text("Some prose text.", use_code_chunker=False)
 
-    patched_chunker_instance.chunk.assert_called_once_with("Some prose text.")
     assert result == ["prose chunk"]

surfsense_backend/tests/unit/indexing_pipeline/test_document_summarizer.py

@@ -0,0 +1,39 @@
+import pytest
+from unittest.mock import MagicMock
+
+from app.indexing_pipeline.document_summarizer import summarize_document
+
+pytestmark = pytest.mark.unit
+
+
+@pytest.mark.usefixtures("patched_summarizer_chain")
+async def test_without_metadata_returns_raw_summary():
+    result = await summarize_document("# Content", llm=MagicMock(model="gpt-4"))
+
+    assert result == "The summary."
+
+
+@pytest.mark.usefixtures("patched_summarizer_chain")
+async def test_with_metadata_includes_metadata_values_in_output():
+    result = await summarize_document(
+        "# Content",
+        llm=MagicMock(model="gpt-4"),
+        metadata={"author": "Alice", "source": "Notion"},
+    )
+
+    assert "Alice" in result
+    assert "Notion" in result
+
+
+@pytest.mark.usefixtures("patched_summarizer_chain")
+async def test_with_metadata_omits_empty_fields_from_output():
+    result = await summarize_document(
+        "# Content",
+        llm=MagicMock(model="gpt-4"),
+        metadata={"author": "Alice", "description": ""},
+    )
+
+    assert "Alice" in result
+    assert "description" not in result.lower()
+
+