VizRef / scripts /test /test_data_utils_comprehensive.py
chenx906's picture
Add model and inference code
18a82fb
Raw
History Blame Contribute Delete
22 kB
# !/usr/bin/env python3
"""
Comprehensive test script for data_utils.py
Tests the orchestration layer that combines transforms, url_dataset, and creates DataLoaders
Run from project root: python test_data_utils_comprehensive.py
"""
import sys
import os
from pathlib import Path
import torch
import numpy as np
from PIL import Image
import json
import tempfile
import traceback
import logging
from typing import Dict, List
from unittest.mock import patch, MagicMock
# Set up logging
logging.basicConfig(level=logging.INFO, format='%(levelname)s: %(message)s')
class TestRunner:
"""Test runner with results tracking"""
def __init__(self):
self.passed = 0
self.failed = 0
self.skipped = 0
self.errors = []
def run_test(self, test_name: str, test_func):
"""Run a single test with error handling"""
print(f"\n{'=' * 60}")
print(f"Running: {test_name}")
print(f"{'=' * 60}")
try:
test_func()
self.passed += 1
print(f"βœ… PASSED: {test_name}")
except FileNotFoundError as e:
self.skipped += 1
print(f"⚠️ SKIPPED: {test_name} - {str(e)}")
except Exception as e:
self.failed += 1
self.errors.append({
'test': test_name,
'error': str(e),
'traceback': traceback.format_exc()
})
print(f"❌ FAILED: {test_name}")
print(f"Error: {str(e)}")
print(f"Traceback:\n{traceback.format_exc()}")
def print_summary(self):
"""Print test summary"""
total = self.passed + self.failed + self.skipped
print(f"\n{'=' * 60}")
print(f"TEST SUMMARY")
print(f"{'=' * 60}")
print(f"Total tests: {total}")
print(f"βœ… Passed: {self.passed}")
print(f"❌ Failed: {self.failed}")
print(f"⚠️ Skipped: {self.skipped}")
if self.errors:
print(f"\nFailed tests:")
for error in self.errors:
print(f" - {error['test']}: {error['error']}")
def create_mock_model_configs():
"""Create mock model configs for testing"""
return {
'efficientnet-b2': {
'input_size': 260,
'batch_size': 32,
},
'resnet50': {
'input_size': 224,
'batch_size': 32,
}
}
def create_mock_split_data(num_items=20, split_name="train"):
"""Create mock dataset split data"""
decades = ['1960s', '1970s', '1980s', '1990s', '2000s']
mock_data = []
for i in range(num_items):
mock_data.append({
"id": f"{split_name}_id_{i}",
"product_id": f"product_{i % 5}", # Some products have multiple images
"name": f"Test Product {i}",
"decade": decades[i % 5], # Ensure all decades represented
"url": f"https://example.com/{split_name}_image_{i}.jpg",
"classification": f"furniture",
"makers": f"Test Maker {i % 3}",
"country": f"Country_{i % 3}"
})
return mock_data
def create_test_environment():
"""Create a complete test environment with mock data"""
temp_dir = tempfile.mkdtemp()
temp_path = Path(temp_dir)
# Create directory structure
data_dir = temp_path / "data"
splits_dir = data_dir / "splits"
cache_dir = data_dir / "cache" / "images"
splits_dir.mkdir(parents=True)
cache_dir.mkdir(parents=True)
# Create split files
split_files = {}
for split_name in ["train", "val", "test"]:
split_data = create_mock_split_data(20 if split_name == "train" else 10, split_name)
split_file = splits_dir / f"{split_name}.json"
with open(split_file, 'w') as f:
json.dump(split_data, f, indent=2)
split_files[split_name] = split_file
return temp_path, data_dir, split_files, cache_dir
def test_imports():
"""Test that all required modules can be imported"""
print("Testing imports...")
try:
# Get project root and add to path
current_dir = Path.cwd()
src_path = current_dir / "src"
if src_path.exists():
sys.path.insert(0, str(current_dir))
from src.data.data_utils import (
create_data_loaders, create_weighted_sampler, analyze_dataset_splits,
create_test_loader, prepare_data_for_training, get_dataset_statistics
)
from src.data.url_dataset import BaseDataset, URLDataset, CachedDataset
from src.data.transforms import get_transforms_for_model
print("βœ“ All data_utils imports successful")
return True
except ImportError as e:
print(f"❌ Import failed: {e}")
return False
def test_create_data_loaders():
"""Test create_data_loaders function"""
print("Testing create_data_loaders...")
# Import after ensuring path is set
from src.data.data_utils import create_data_loaders
from torchvision import transforms
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Mock the model configs to avoid import issues
with patch('src.data.transforms.TRAINING_CONFIGS', create_mock_model_configs()):
# Create test config
config = {
'model_name': 'efficientnet-b2',
'batch_size': 4,
'num_workers': 0, # Use 0 for testing
'data_dir': str(data_dir),
'use_cached': False,
'use_class_weights': False,
'use_weighted_sampling': False,
'augmentation_level': 'medium',
'max_download_retries': 1,
'download_timeout': 5
}
# Mock image downloads to avoid network calls
mock_image = Image.fromarray(np.random.randint(0, 255, (200, 200, 3), dtype=np.uint8))
def mock_load_image(self, item):
return mock_image
with patch('src.data.url_dataset.URLDataset._load_image', mock_load_image):
# Test basic data loader creation
train_loader, val_loader, class_weights, class_names = create_data_loaders(
config,
data_dir=data_dir,
use_subset=True,
subset_fraction=0.5 # Use 50% for testing
)
print(f"βœ“ Created DataLoaders successfully")
print(f" Train loader: {len(train_loader)} batches")
print(f" Val loader: {len(val_loader)} batches")
print(f" Class names: {class_names}")
# Validate results
assert len(train_loader) > 0, "Train loader should have batches"
assert len(val_loader) > 0, "Val loader should have batches"
assert class_names == ['1960s', '1970s', '1980s', '1990s', '2000s'], "Wrong class names"
assert class_weights is None, "Class weights should be None when disabled"
# Test loading a batch
for batch_idx, batch_data in enumerate(train_loader):
if len(batch_data) == 3:
images, labels, metadata = batch_data
else:
images, labels = batch_data[0], batch_data[1]
metadata = None
print(f"βœ“ Loaded batch: images={images.shape}, labels={labels.shape}")
# Validate batch
assert images.shape[0] <= config['batch_size'], "Batch size exceeded"
assert images.shape[1] == 3, "Wrong number of channels"
assert labels.shape[0] == images.shape[0], "Label count mismatch"
break # Just test first batch
finally:
# Cleanup
import shutil
shutil.rmtree(temp_path)
def test_weighted_sampler():
"""Test create_weighted_sampler function"""
print("Testing create_weighted_sampler...")
from src.data.data_utils import create_weighted_sampler
from src.data.url_dataset import BaseDataset
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Create dataset with imbalanced data
imbalanced_data = []
decades = ['1960s', '1970s', '1980s', '1990s', '2000s']
# Create imbalanced distribution
counts = [20, 5, 10, 3, 15] # Imbalanced
item_id = 0
for decade_idx, count in enumerate(counts):
for i in range(count):
imbalanced_data.append({
"id": f"id_{item_id}",
"product_id": f"product_{item_id}",
"name": f"Product {item_id}",
"decade": decades[decade_idx],
"url": f"https://example.com/image_{item_id}.jpg",
"classification": "test",
"makers": "test",
"country": "test"
})
item_id += 1
# Save imbalanced split
imbalanced_split = data_dir / "splits" / "imbalanced.json"
with open(imbalanced_split, 'w') as f:
json.dump(imbalanced_data, f)
# Create dataset
dataset = BaseDataset(str(imbalanced_split))
print(f"βœ“ Created imbalanced dataset with {len(dataset)} samples")
# Test label distribution before sampling
labels = dataset.get_labels()
from collections import Counter
original_counts = Counter(labels)
print(f" Original distribution: {dict(original_counts)}")
# Create weighted sampler
sampler = create_weighted_sampler(dataset)
print(f"βœ“ Created weighted sampler with {len(sampler)} samples")
# Test sampling
sampled_indices = list(sampler)[:100] # Sample 100 items
sampled_labels = [labels[idx] for idx in sampled_indices]
sampled_counts = Counter(sampled_labels)
print(f" Sampled distribution: {dict(sampled_counts)}")
# Check if sampling is more balanced
original_ratio = max(original_counts.values()) / min(original_counts.values())
sampled_ratio = max(sampled_counts.values()) / min(sampled_counts.values()) if min(
sampled_counts.values()) > 0 else float('inf')
print(f" Balance improvement: {original_ratio:.2f} -> {sampled_ratio:.2f}")
# Sampled distribution should be more balanced
assert sampled_ratio < original_ratio or sampled_ratio < 3.0, "Sampling should improve balance"
finally:
import shutil
shutil.rmtree(temp_path)
def test_analyze_dataset_splits():
"""Test analyze_dataset_splits function"""
print("Testing analyze_dataset_splits...")
from src.data.data_utils import analyze_dataset_splits
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Analyze the splits
analysis = analyze_dataset_splits(data_dir)
print(f"βœ“ Analysis completed for {len(analysis)} splits")
# Check analysis structure
expected_splits = ['train', 'val', 'test']
for split_name in expected_splits:
assert split_name in analysis, f"Missing analysis for {split_name}"
split_stats = analysis[split_name]
print(f" {split_name}: {split_stats['total_images']} images, {split_stats['unique_products']} products")
# Validate structure
required_keys = ['total_images', 'unique_products', 'decades', 'classifications', 'countries']
for key in required_keys:
assert key in split_stats, f"Missing key {key} in {split_name} analysis"
# Check decades distribution
decades_dist = split_stats['decades']
assert len(decades_dist) == 5, f"Should have 5 decades, got {len(decades_dist)}"
# Check for data leakage detection
# Our test data has overlapping product_ids, so leakage should be detected
if 'data_leakage' in analysis:
print(f" ⚠️ Data leakage detected: {analysis['data_leakage']['train_val_overlap']} overlapping products")
else:
print(f" βœ“ No data leakage detected")
finally:
import shutil
shutil.rmtree(temp_path)
def test_prepare_data_for_training():
"""Test prepare_data_for_training function"""
print("Testing prepare_data_for_training...")
from src.data.data_utils import prepare_data_for_training
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Test config
config = {
'data_dir': str(data_dir),
'use_cached': False
}
# FIXED: Test with data leakage detection disabled first
is_ready = prepare_data_for_training(
config,
download_if_missing=False,
verify_splits=False # Disable verification to test basic functionality
)
print(f"βœ“ Data preparation (no verification): {is_ready}")
assert is_ready == True, "Data preparation should succeed when verification disabled"
# FIXED: Test with verification enabled (should detect leakage)
is_ready_with_verification = prepare_data_for_training(
config,
download_if_missing=False,
verify_splits=True # Enable verification
)
print(f"βœ“ Data preparation (with verification): {is_ready_with_verification}")
# This should be False because our mock data has intentional leakage
assert is_ready_with_verification == False, "Data preparation should fail when data leakage detected"
# FIXED: Test with missing split files
missing_split = data_dir / "splits" / "train.json"
missing_split.unlink() # Remove train split
is_ready_missing = prepare_data_for_training(
config,
download_if_missing=False,
verify_splits=False
)
print(f"βœ“ Data preparation with missing files: {is_ready_missing}")
assert is_ready_missing == False, "Data preparation should fail with missing splits"
print("βœ“ All data preparation scenarios tested correctly")
finally:
import shutil
shutil.rmtree(temp_path)
def test_get_dataset_statistics():
"""Test get_dataset_statistics function"""
print("Testing get_dataset_statistics...")
from src.data.data_utils import get_dataset_statistics
from src.data.url_dataset import BaseDataset
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Create dataset
dataset = BaseDataset(str(split_files['train']))
# Get statistics
stats = get_dataset_statistics(dataset)
print(f"βœ“ Generated statistics for dataset")
# Validate statistics structure
required_keys = [
'total_samples', 'num_classes', 'class_names',
'class_distribution', 'class_balance', 'product_stats',
'top_classifications', 'top_countries'
]
for key in required_keys:
assert key in stats, f"Missing statistics key: {key}"
print(f" Total samples: {stats['total_samples']}")
print(f" Number of classes: {stats['num_classes']}")
print(f" Class distribution: {stats['class_distribution']}")
print(f" Imbalance ratio: {stats['class_balance']['imbalance_ratio']:.2f}")
print(f" Unique products: {stats['product_stats']['unique_products']}")
print(f" Avg images per product: {stats['product_stats']['avg_images_per_product']:.2f}")
# Validate values
assert stats['total_samples'] == len(dataset), "Wrong total samples count"
assert stats['num_classes'] == 5, "Wrong number of classes"
assert len(stats['class_names']) == 5, "Wrong number of class names"
finally:
import shutil
shutil.rmtree(temp_path)
def test_create_test_loader():
"""Test create_test_loader function"""
print("Testing create_test_loader...")
from src.data.data_utils import create_test_loader
temp_path, data_dir, split_files, cache_dir = create_test_environment()
try:
# Mock the model configs
with patch('src.data.transforms.TRAINING_CONFIGS', create_mock_model_configs()):
config = {
'model_name': 'resnet50',
'batch_size': 4,
'num_workers': 0,
'data_dir': str(data_dir),
'use_cached': False,
'max_download_retries': 1,
'download_timeout': 5
}
# Mock image loading
mock_image = Image.fromarray(np.random.randint(0, 255, (200, 200, 3), dtype=np.uint8))
def mock_load_image(self, item):
return mock_image
with patch('src.data.url_dataset.URLDataset._load_image', mock_load_image):
test_loader, class_names = create_test_loader(
config,
data_dir=data_dir,
batch_size=6
)
print(f"βœ“ Created test loader with {len(test_loader)} batches")
print(f" Class names: {class_names}")
# Validate
assert len(test_loader) > 0, "Test loader should have batches"
assert class_names == ['1960s', '1970s', '1980s', '1990s', '2000s'], "Wrong class names"
# Test loading a batch
for batch_data in test_loader:
if len(batch_data) == 3:
images, labels, metadata = batch_data
else:
images, labels = batch_data[0], batch_data[1]
print(f"βœ“ Test batch: images={images.shape}, labels={labels.shape}")
assert images.shape[0] <= 6, "Batch size should respect config"
break
finally:
import shutil
shutil.rmtree(temp_path)
def test_integration_with_real_data():
"""Test integration with real project data if available"""
print("Testing integration with real data...")
# Check if real data exists
project_root = Path.cwd()
real_data_dir = project_root / "data"
real_splits_dir = real_data_dir / "splits"
real_train_split = real_splits_dir / "train.json"
if not real_train_split.exists():
print("⚠️ Real data not found, skipping integration test")
raise FileNotFoundError("Real data not available")
from src.data.data_utils import create_data_loaders, analyze_dataset_splits
try:
# Mock the model configs
with patch('src.data.transforms.TRAINING_CONFIGS', create_mock_model_configs()):
print(f"βœ“ Found real data at {real_data_dir}")
# Test analysis with real data
analysis = analyze_dataset_splits(real_data_dir)
for split_name, stats in analysis.items():
if isinstance(stats, dict) and 'total_images' in stats:
print(f" {split_name}: {stats['total_images']} images")
# Test data loader creation with real data (small subset)
config = {
'model_name': 'efficientnet-b2',
'batch_size': 4,
'num_workers': 0,
'data_dir': str(real_data_dir),
'use_cached': False,
'use_class_weights': False,
'use_weighted_sampling': False,
'augmentation_level': 'light',
'max_download_retries': 1,
'download_timeout': 5
}
# Use very small subset for quick testing
train_loader, val_loader, class_weights, class_names = create_data_loaders(
config,
data_dir=real_data_dir,
use_subset=True,
subset_fraction=0.001 # 0.1% for very quick test
)
print(f"βœ“ Real data integration successful")
print(f" Train batches: {len(train_loader)}")
print(f" Val batches: {len(val_loader)}")
except Exception as e:
print(f"⚠️ Real data integration test failed: {e}")
raise
def main():
"""Main test function"""
print("DATA_UTILS.PY COMPREHENSIVE TEST SUITE")
print("=" * 60)
print("Testing data orchestration and integration layer")
runner = TestRunner()
# Check imports first
if not test_imports():
print("❌ Cannot proceed without successful imports")
return 1
# Define all tests
tests = [
("Create Data Loaders", test_create_data_loaders),
("Weighted Sampler", test_weighted_sampler),
("Analyze Dataset Splits", test_analyze_dataset_splits),
("Prepare Data for Training", test_prepare_data_for_training),
("Get Dataset Statistics", test_get_dataset_statistics),
("Create Test Loader", test_create_test_loader),
("Integration with Real Data", test_integration_with_real_data),
]
# Run all tests
for test_name, test_func in tests:
runner.run_test(test_name, test_func)
# Print summary
runner.print_summary()
# Return exit code
success = runner.failed == 0
if success:
print("\nπŸŽ‰ ALL DATA_UTILS TESTS PASSED!")
print("βœ… The complete data pipeline is working correctly!")
print("\nYour data loading system is ready for:")
print(" πŸš€ Model training")
print(" πŸ“Š Data analysis")
print(" πŸ”„ Production deployment")
else:
print("\n❌ SOME TESTS FAILED!")
print("Check the issues above before proceeding to training")
return 0 if success else 1
if __name__ == "__main__":
exit_code = main()
print(f"\nTest completed with exit code: {exit_code}")