"""Provider tests for DuplicateDetectorProvider.""" from __future__ import annotations import cv2 import numpy as np import pytest from config.settings import Settings from pipeline.feature_extraction import PipelineOutput from providers.forensics.duplicate_detector import DuplicateDetectorProvider @pytest.fixture def duplicate_provider(): return DuplicateDetectorProvider(settings=Settings(environment="test", db_path=":memory:")) def _make_pipeline(image: np.ndarray, original_bytes: bytes | None = None) -> PipelineOutput: return PipelineOutput( image=image, image_hash="h", width=image.shape[1], height=image.shape[0], source="bytes", original_bytes=original_bytes, original_format=".jpg", ) class TestDuplicateDetectorProvider: def test_name(self, duplicate_provider): assert duplicate_provider.name == "duplicate_detector" def test_capability(self, duplicate_provider): from models.providers import ProviderCapability assert duplicate_provider.capability == ProviderCapability.FORENSICS def test_is_available(self, duplicate_provider): assert duplicate_provider.is_available() is True def test_first_image_not_duplicate(self, duplicate_provider, sample_image_bytes): img = cv2.imdecode(np.frombuffer(sample_image_bytes, np.uint8), cv2.IMREAD_COLOR) po = _make_pipeline(img, original_bytes=sample_image_bytes) result = duplicate_provider.execute(po) assert result.success is True assert result.normalized["is_duplicate"] is False assert result.normalized["duplicate_of"] is None # Hashes should be present — 64 bits = 16 hex chars assert "phash" in result.normalized["details"] assert "dhash" in result.normalized["details"] assert len(result.normalized["details"]["phash"]) == 16 # 64 bits as hex def test_identical_image_detected_as_duplicate(self, duplicate_provider, sample_image_bytes): img = cv2.imdecode(np.frombuffer(sample_image_bytes, np.uint8), cv2.IMREAD_COLOR) po1 = _make_pipeline(img, original_bytes=sample_image_bytes) po2 = _make_pipeline(img.copy(), original_bytes=sample_image_bytes) # First call registers duplicate_provider.execute(po1) # Second call should detect duplicate result = duplicate_provider.execute(po2) assert result.normalized["is_duplicate"] is True def test_different_image_not_duplicate(self, duplicate_provider, sample_image_bytes, sample_face_image_bytes): img1 = cv2.imdecode(np.frombuffer(sample_image_bytes, np.uint8), cv2.IMREAD_COLOR) img2 = cv2.imdecode(np.frombuffer(sample_face_image_bytes, np.uint8), cv2.IMREAD_COLOR) po1 = _make_pipeline(img1, original_bytes=sample_image_bytes) po2 = _make_pipeline(img2, original_bytes=sample_face_image_bytes) duplicate_provider.execute(po1) result = duplicate_provider.execute(po2) assert result.normalized["is_duplicate"] is False def test_phash_stable_for_same_image(self, duplicate_provider, sample_image_bytes): img = cv2.imdecode(np.frombuffer(sample_image_bytes, np.uint8), cv2.IMREAD_COLOR) po1 = _make_pipeline(img, original_bytes=sample_image_bytes) po2 = _make_pipeline(img.copy(), original_bytes=sample_image_bytes) r1 = duplicate_provider.execute(po1) r2 = duplicate_provider.execute(po2) # phash should be the same for identical images assert r1.normalized["details"]["phash"] == r2.normalized["details"]["phash"]