from unittest.mock import MagicMock
import pytest
from src.core.settings import ImageCaptionerSettings, Settings
from src.ingestion.models import Chunk
from src.ingestion.transform.image_captioner import ImageCaptioner
from src.libs.llm.base_llm import BaseLLM
class MockLLM(BaseLLM):
def chat(self, messages, **kwargs):
return "A description of the image."
@pytest.fixture
def mock_settings():
settings = MagicMock(spec=Settings)
# Setup nested structure mocks
settings.ingestion = MagicMock()
settings.ingestion.transform = MagicMock()
settings.ingestion.transform.image_captioner = ImageCaptionerSettings(
enabled=True,
fallback_on_error=True,
prompt_path="config/prompts/image_captioning.txt",
)
settings.vision_llm = MagicMock()
settings.vision_llm.provider = "openai"
settings.llm = MagicMock()
settings.llm.api_key = "dummy"
return settings
def test_image_captioning_success(mock_settings):
"""Test successful caption generation when enabled."""
mock_llm = MockLLM()
chunk = Chunk(text="Test chunk", metadata={"image_refs": ["img1"]})
captioner = ImageCaptioner(mock_settings, llm=mock_llm)
results = captioner.transform([chunk])
assert len(results) == 1
assert "image_captions" in results[0].metadata
assert (
results[0].metadata["image_captions"]["img1"] == "A description of the image."
)
def test_image_captioning_fallback(mock_settings):
"""Test fallback behavior when LLM fails."""
mock_llm = MagicMock(spec=BaseLLM)
mock_llm.chat.side_effect = RuntimeError("API Error")
chunk = Chunk(text="Test chunk", metadata={"image_refs": ["img1"]})
captioner = ImageCaptioner(mock_settings, llm=mock_llm)
results = captioner.transform([chunk])
assert len(results) == 1
assert "image_captions" not in results[0].metadata
assert results[0].metadata.get("has_unprocessed_images") is True
errors = results[0].metadata.get("processing_errors")
assert errors is not None
assert any("API Error" in str(e) for e in errors)
def test_image_captioning_disabled(mock_settings):
"""Test that nothing happens when disabled."""
mock_settings.ingestion.transform.image_captioner.enabled = False
# Even with a valid LLM, it should skip
mock_llm = MockLLM()
chunk = Chunk(text="Test chunk", metadata={"image_refs": ["img1"]})
captioner = ImageCaptioner(mock_settings, llm=mock_llm)
results = captioner.transform([chunk])
assert "image_captions" not in results[0].metadata
def test_no_image_refs(mock_settings):
"""Test that chunks without images are ignored."""
mock_llm = MockLLM()
chunk = Chunk(text="Test chunk", metadata={}) # No image_refs
captioner = ImageCaptioner(mock_settings, llm=mock_llm)
results = captioner.transform([chunk])
assert "image_captions" not in results[0].metadata
assert "has_unprocessed_images" not in results[0].metadata
