"""
Example scripts for Synthetic Dataset Generator
Demonstrates various dataset generation scenarios
"""

from dataset_generator import SyntheticDatasetGenerator
import os


def example_1_simple_image_classification():
    """Example 1: Generate a simple image classification dataset"""
    print("\n" + "=" * 60)
    print("Example 1: Simple Image Classification Dataset")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output",
        use_cuda=True
    )

    # Generate dataset with 3 classes, 10 samples each
    result = generator.generate_image_classification_dataset(
        classes=["apple", "orange", "banana"],
        samples_per_class=10,
        image_size=(512, 512),
        style="photorealistic",
        guidance_scale=7.5,
        num_inference_steps=30  # Reduced for speed
    )

    print(f"\n✅ Dataset created at: {result['dataset_path']}")
    print(f"   Total images: {result['labels_count']}")
    print(f"   Classes: {result['metadata']['classes']}")


def example_2_object_detection():
    """Example 2: Generate object detection dataset with COCO format"""
    print("\n" + "=" * 60)
    print("Example 2: Object Detection Dataset (COCO format)")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    # Generate dataset with multiple objects per image
    result = generator.generate_object_detection_dataset(
        objects=["person", "car", "dog"],
        num_images=20,
        objects_per_image=(1, 3),
        image_size=(1024, 1024),
        annotation_format="coco"
    )

    print(f"\n✅ Dataset created at: {result['dataset_path']}")
    print(f"   Format: COCO JSON")
    print(f"   Object classes: {result['metadata']['objects']}")


def example_3_text_classification():
    """Example 3: Generate text classification dataset for sentiment analysis"""
    print("\n" + "=" * 60)
    print("Example 3: Text Classification (Sentiment Analysis)")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    # Generate sentiment dataset
    result = generator.generate_text_classification_dataset(
        categories=["positive", "negative", "neutral"],
        samples_per_category=50,
        text_type="sentence",
        min_length=10,
        max_length=50
    )

    print(f"\n✅ Dataset created at: {result['dataset_path']}")
    print(f"   Total samples: {result['samples_count']}")
    print(f"   Categories: {result['metadata']['categories']}")
    print(f"   Splits: train.json, validation.json, test.json")


def example_4_ner_dataset():
    """Example 4: Generate Named Entity Recognition dataset"""
    print("\n" + "=" * 60)
    print("Example 4: Named Entity Recognition (NER)")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    # Generate NER dataset
    result = generator.generate_ner_dataset(
        entity_types=["PERSON", "ORGANIZATION", "LOCATION"],
        num_samples=100,
        entities_per_sample=(2, 4)
    )

    print(f"\n✅ Dataset created at: {result['dataset_path']}")
    print(f"   Total samples: {result['samples_count']}")
    print(f"   Entity types: {result['metadata']['entity_types']}")


def example_5_code_generation():
    """Example 5: Generate code dataset for bug detection"""
    print("\n" + "=" * 60)
    print("Example 5: Code Generation (Bug Detection)")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    # Generate code dataset
    result = generator.generate_code_dataset(
        task_type="bug_detection",
        programming_languages=["python", "javascript"],
        num_samples=50
    )

    print(f"\n✅ Dataset created at: {result['dataset_path']}")
    print(f"   Total samples: {result['samples_count']}")
    print(f"   Languages: {result['metadata']['languages']}")
    print(f"   Task: {result['metadata']['dataset_type']}")


def example_6_custom_style_images():
    """Example 6: Generate images with different artistic styles"""
    print("\n" + "=" * 60)
    print("Example 6: Custom Artistic Styles")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    styles = ["photorealistic", "cartoon", "3d render", "artistic"]

    for style in styles:
        print(f"\nGenerating {style} style images...")

        result = generator.generate_image_classification_dataset(
            classes=["landscape"],
            samples_per_class=5,
            image_size=(512, 512),
            style=style,
            guidance_scale=8.0,
            num_inference_steps=30
        )

        print(f"   ✅ {style}: {result['dataset_path']}")


def example_7_large_scale_dataset():
    """Example 7: Generate a large-scale dataset efficiently"""
    print("\n" + "=" * 60)
    print("Example 7: Large-Scale Dataset Generation")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output",
        use_cuda=True,
        use_fp16=True  # Memory optimization
    )

    # Generate large dataset (adjust numbers based on your hardware)
    result = generator.generate_image_classification_dataset(
        classes=["cat", "dog", "bird", "car", "tree", "building"],
        samples_per_class=100,  # 600 total images
        image_size=(512, 512),  # Smaller size for speed
        style="photorealistic",
        guidance_scale=7.0,
        num_inference_steps=30  # Faster generation
    )

    print(f"\n✅ Large dataset created: {result['dataset_path']}")
    print(f"   Total images: {result['labels_count']}")
    print(f"   Estimated time: ~30-60 minutes on RTX 4090")


def example_8_yolo_format():
    """Example 8: Generate object detection in YOLO format"""
    print("\n" + "=" * 60)
    print("Example 8: Object Detection (YOLO format)")
    print("=" * 60 + "\n")

    generator = SyntheticDatasetGenerator(
        output_dir="./examples_output"
    )

    result = generator.generate_object_detection_dataset(
        objects=["person", "bicycle", "car", "motorcycle"],
        num_images=50,
        objects_per_image=(1, 5),
        image_size=(640, 640),  # Common YOLO input size
        annotation_format="yolo"
    )

    print(f"\n✅ YOLO dataset created at: {result['dataset_path']}")
    print(f"   Each image has a corresponding .txt annotation file")
    print(f"   Format: class_id x_center y_center width height (normalized)")


def run_all_examples():
    """Run all examples sequentially"""
    print("\n" + "=" * 60)
    print("Running All Examples")
    print("=" * 60)
    print("\nNOTE: This will take a while and requires GPU!")
    print("Press Ctrl+C to cancel\n")

    try:
        # Quick examples (no heavy model loading)
        example_3_text_classification()
        example_4_ner_dataset()
        example_5_code_generation()

        # Image examples (requires GPU)
        response = input("\nRun image generation examples? (y/n): ")
        if response.lower() == 'y':
            example_1_simple_image_classification()
            example_2_object_detection()
            example_6_custom_style_images()
            example_8_yolo_format()

            response = input("\nRun large-scale example? (600 images, ~30-60 min): (y/n): ")
            if response.lower() == 'y':
                example_7_large_scale_dataset()

        print("\n" + "=" * 60)
        print("✅ All examples completed!")
        print("=" * 60)

    except KeyboardInterrupt:
        print("\n\n❌ Examples cancelled by user")
    except Exception as e:
        print(f"\n\n❌ Error: {e}")


def main():
    """Main menu for running examples"""
    print("\n" + "=" * 60)
    print("Synthetic Dataset Generator - Examples")
    print("=" * 60)
    print("\nAvailable examples:")
    print("  1. Simple Image Classification")
    print("  2. Object Detection (COCO)")
    print("  3. Text Classification (Sentiment)")
    print("  4. Named Entity Recognition")
    print("  5. Code Generation (Bug Detection)")
    print("  6. Custom Artistic Styles")
    print("  7. Large-Scale Dataset")
    print("  8. Object Detection (YOLO)")
    print("  9. Run all examples")
    print("  0. Exit")

    while True:
        try:
            choice = input("\nSelect example (0-9): ").strip()

            if choice == '0':
                print("Goodbye!")
                break
            elif choice == '1':
                example_1_simple_image_classification()
            elif choice == '2':
                example_2_object_detection()
            elif choice == '3':
                example_3_text_classification()
            elif choice == '4':
                example_4_ner_dataset()
            elif choice == '5':
                example_5_code_generation()
            elif choice == '6':
                example_6_custom_style_images()
            elif choice == '7':
                example_7_large_scale_dataset()
            elif choice == '8':
                example_8_yolo_format()
            elif choice == '9':
                run_all_examples()
            else:
                print("Invalid choice. Please select 0-9.")

        except KeyboardInterrupt:
            print("\n\nGoodbye!")
            break
        except Exception as e:
            print(f"\n❌ Error: {e}")
            print("Please try again or select a different example.")


if __name__ == "__main__":
    main()