gmakstutis/cognee
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Test: Mock file download and open in musique adapter (#571)

Browse source 
<!-- .github/pull_request_template.md -->

## Description
<!-- Provide a clear description of the changes in this PR -->

## DCO Affirmation
I affirm that all code in every commit of this pull request conforms to
the terms of the Topoteretes Developer Certificate of Origin


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

- **Tests**
- Enhanced test coverage to improve adapter instantiation and data
loading reliability.
  - Updated mock testing logic to ensure robust content handling.
  - Removed an outdated test focused on data limit validation.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
This commit is contained in:
alekszievr 2025-02-20 16:11:19 +01:00 • committed by GitHub
parent 97db017708
commit 28f92f661e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 34 additions and 23 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

57
cognee/tests/unit/eval_framework/benchmark_adapters_test.py
View file

@ -3,6 +3,13 @@ from evals.eval_framework.benchmark_adapters.hotpot_qa_adapter import HotpotQAAd
from evals.eval_framework.benchmark_adapters.musique_adapter import MusiqueQAAdapter
from evals.eval_framework.benchmark_adapters.dummy_adapter import DummyAdapter
from evals.eval_framework.benchmark_adapters.twowikimultihop_adapter import TwoWikiMultihopAdapter
from unittest.mock import patch, mock_open
MOCK_JSONL_DATA = """\
{"id": "1", "question": "What is AI?", "answer": "Artificial Intelligence", "paragraphs": [{"paragraph_text": "AI is a field of computer science."}]}
{"id": "2", "question": "What is ML?", "answer": "Machine Learning", "paragraphs": [{"paragraph_text": "ML is a subset of AI."}]}
"""
ADAPTER_CLASSES = [
@ -19,8 +26,18 @@ def test_adapter_can_instantiate_and_load(AdapterClass):
Basic smoke test: instantiate each adapter, call load_corpus with no limit,
and ensure it returns the expected tuple of (list, list).
"""
adapter = AdapterClass()
result = adapter.load_corpus()
if AdapterClass == MusiqueQAAdapter:
with (
patch.object(MusiqueQAAdapter, "_musique_download_file"),
patch("builtins.open", new_callable=mock_open, read_data=MOCK_JSONL_DATA),
patch("os.path.exists", return_value=True),
):
adapter = AdapterClass()
result = adapter.load_corpus()
else:
adapter = AdapterClass()
result = adapter.load_corpus()
assert isinstance(result, tuple), f"{AdapterClass.__name__} did not return a tuple."
assert len(result) == 2, f"{AdapterClass.__name__} returned tuple of length != 2."
@ -38,32 +55,26 @@ def test_adapter_returns_some_content(AdapterClass):
Verify that the adapter returns some data and that each QA dict
at least has a 'question' and 'answer' key (you can extend or remove as needed).
"""
adapter = AdapterClass()
limit = 3
if AdapterClass == MusiqueQAAdapter:
with (
patch.object(MusiqueQAAdapter, "_musique_download_file"),
patch("builtins.open", new_callable=mock_open, read_data=MOCK_JSONL_DATA),
patch("os.path.exists", return_value=True),
):
adapter = AdapterClass()
corpus_list, qa_pairs = adapter.load_corpus(limit=limit)
else:
adapter = AdapterClass()
corpus_list, qa_pairs = adapter.load_corpus(limit=limit)
corpus_list, qa_pairs = adapter.load_corpus(limit=3) # small limit
# We don't know how large the dataset is, but we expect at least 1 item
assert len(corpus_list) > 0, f"{AdapterClass.__name__} returned an empty corpus_list."
assert len(qa_pairs) > 0, f"{AdapterClass.__name__} returned an empty question_answer_pairs."
assert len(qa_pairs) <= limit, (
f"{AdapterClass.__name__} returned more QA items than requested limit={limit}."
)
for item in qa_pairs:
assert "question" in item, f"{AdapterClass.__name__} missing 'question' key in QA pair."
assert "answer" in item, f"{AdapterClass.__name__} missing 'answer' key in QA pair."
@pytest.mark.parametrize("AdapterClass", ADAPTER_CLASSES)
def test_adapter_limit(AdapterClass):
"""
Check that the `limit` parameter correctly restricts the amount of data returned.
We'll test with limit=5.
"""
adapter = AdapterClass()
limit = 5
corpus_list, qa_pairs = adapter.load_corpus(limit=limit)
# Confirm that we didn't receive more than 'limit'
# (Some adapters might be allowed to return fewer if the dataset is small)
assert len(qa_pairs) <= limit, (
f"{AdapterClass.__name__} returned more QA items than requested limit={limit}."
)