API_CLIENT_EXAMPLES.py

"""
A/B Test Predictor - API Client Examples
==========================================

This file shows how to send requests to the A/B Test Predictor API.
"""

# ============================================================================
# Option 1: Gradio Python Client (Recommended)
# ============================================================================

from gradio_client import Client
from PIL import Image
import json

# Initialize the client
# For local deployment:
client = Client("http://localhost:7860")

# For Hugging Face Spaces deployment:
# client = Client("your-username/ABTestPredictor")

def predict_with_gradio_client(control_image_path, variant_image_path, 
                                business_model, customer_type, conversion_type, 
                                industry, page_type):
    """
    Send prediction request using Gradio Client
    
    Args:
        control_image_path: Path to control image file
        variant_image_path: Path to variant image file
        business_model: One of ["E-Commerce", "Lead Generation", "Other*", "SaaS"]
        customer_type: One of ["B2B", "B2C", "Both", "Other*"]
        conversion_type: One of ["Direct Purchase", "High-Intent Lead Gen", 
                                 "Info/Content Lead Gen", "Location Search", 
                                 "Non-Profit/Community", "Other Conversion"]
        industry: One of the 14 industry categories
        page_type: One of ["Awareness & Discovery", "Consideration & Evaluation", 
                           "Conversion", "Internal & Navigation", "Post-Conversion & Other"]
    
    Returns:
        dict: Prediction results with confidence scores
    """
    result = client.predict(
        control_image_path,      # Control image file path
        variant_image_path,      # Variant image file path
        business_model,          # Business Model dropdown
        customer_type,           # Customer Type dropdown
        conversion_type,         # Conversion Type dropdown
        industry,                # Industry dropdown
        page_type,               # Page Type dropdown
        api_name="/predict_with_categorical_data"  # The function endpoint
    )
    
    return result


# Example usage
if __name__ == "__main__":
    # Example 1: Basic prediction
    result = predict_with_gradio_client(
        control_image_path="path/to/control_image.jpg",
        variant_image_path="path/to/variant_image.jpg",
        business_model="SaaS",
        customer_type="B2B",
        conversion_type="High-Intent Lead Gen",
        industry="B2B Software & Tech",
        page_type="Awareness & Discovery"
    )
    
    print("Prediction Results:")
    print(json.dumps(result, indent=2))
    
    # Access specific fields
    win_probability = result['predictionResults']['probability']
    confidence = result['predictionResults']['modelConfidence']
    
    print(f"\nWin Probability: {win_probability}")
    print(f"Model Confidence: {confidence}%")


# ============================================================================
# Option 2: Direct HTTP POST Request (cURL equivalent in Python)
# ============================================================================

import requests
import base64

def predict_with_http_request(control_image_path, variant_image_path,
                               business_model, customer_type, conversion_type,
                               industry, page_type, api_url="http://localhost:7860"):
    """
    Send prediction request using direct HTTP POST
    
    Note: This requires converting images to base64 for Gradio's API format
    """
    
    # Read and encode images
    with open(control_image_path, "rb") as f:
        control_b64 = base64.b64encode(f.read()).decode()
    
    with open(variant_image_path, "rb") as f:
        variant_b64 = base64.b64encode(f.read()).decode()
    
    # Prepare the request payload (Gradio format)
    payload = {
        "data": [
            f"data:image/jpeg;base64,{control_b64}",  # Control image
            f"data:image/jpeg;base64,{variant_b64}",  # Variant image
            business_model,
            customer_type,
            conversion_type,
            industry,
            page_type
        ]
    }
    
    # Send POST request to Gradio API
    response = requests.post(
        f"{api_url}/api/predict",
        json=payload,
        headers={"Content-Type": "application/json"}
    )
    
    if response.status_code == 200:
        return response.json()['data'][0]  # Gradio wraps response in 'data' array
    else:
        raise Exception(f"API request failed: {response.status_code} - {response.text}")


# ============================================================================
# Option 3: Using PIL Images (in-memory)
# ============================================================================

import numpy as np
from PIL import Image

def predict_with_pil_images(control_img, variant_img,
                             business_model, customer_type, conversion_type,
                             industry, page_type):
    """
    Send prediction with PIL Image objects (useful for programmatic image generation)
    
    Args:
        control_img: PIL Image object
        variant_img: PIL Image object
    """
    
    # Convert PIL images to numpy arrays (Gradio expects numpy arrays)
    control_array = np.array(control_img)
    variant_array = np.array(variant_img)
    
    # Use the Gradio client
    result = client.predict(
        control_array,
        variant_array,
        business_model,
        customer_type,
        conversion_type,
        industry,
        page_type,
        api_name="/predict_with_categorical_data"
    )
    
    return result


# Example with PIL
if __name__ == "__main__":
    # Load images using PIL
    control_img = Image.open("control.jpg")
    variant_img = Image.open("variant.jpg")
    
    result = predict_with_pil_images(
        control_img=control_img,
        variant_img=variant_img,
        business_model="SaaS",
        customer_type="B2B",
        conversion_type="High-Intent Lead Gen",
        industry="B2B Software & Tech",
        page_type="Awareness & Discovery"
    )


# ============================================================================
# Option 4: Batch Processing Multiple Tests
# ============================================================================

def batch_predict(test_cases, output_file="results.json"):
    """
    Process multiple A/B tests in batch
    
    Args:
        test_cases: List of dicts with test parameters
        output_file: Where to save results
    
    Example test_cases:
        [
            {
                "control_image": "test1_control.jpg",
                "variant_image": "test1_variant.jpg",
                "business_model": "SaaS",
                "customer_type": "B2B",
                "conversion_type": "High-Intent Lead Gen",
                "industry": "B2B Software & Tech",
                "page_type": "Awareness & Discovery"
            },
            # ... more tests
        ]
    """
    
    results = []
    
    for i, test in enumerate(test_cases):
        print(f"Processing test {i+1}/{len(test_cases)}...")
        
        try:
            result = predict_with_gradio_client(
                control_image_path=test["control_image"],
                variant_image_path=test["variant_image"],
                business_model=test["business_model"],
                customer_type=test["customer_type"],
                conversion_type=test["conversion_type"],
                industry=test["industry"],
                page_type=test["page_type"]
            )
            
            results.append({
                "test_id": i + 1,
                "input": test,
                "prediction": result
            })
            
        except Exception as e:
            print(f"Error processing test {i+1}: {e}")
            results.append({
                "test_id": i + 1,
                "input": test,
                "error": str(e)
            })
    
    # Save results
    with open(output_file, "w") as f:
        json.dump(results, f, indent=2)
    
    print(f"\nBatch processing complete! Results saved to {output_file}")
    return results


# ============================================================================
# Valid Category Values (for reference)
# ============================================================================

VALID_CATEGORIES = {
    "business_model": [
        "E-Commerce",
        "Lead Generation",
        "Other*",
        "SaaS"
    ],
    
    "customer_type": [
        "B2B",
        "B2C",
        "Both",
        "Other*"
    ],
    
    "conversion_type": [
        "Direct Purchase",
        "High-Intent Lead Gen",
        "Info/Content Lead Gen",
        "Location Search",
        "Non-Profit/Community",
        "Other Conversion"
    ],
    
    "industry": [
        "Automotive & Transportation",
        "B2B Services",
        "B2B Software & Tech",
        "Consumer Services",
        "Consumer Software & Apps",
        "Education",
        "Finance, Insurance & Real Estate",
        "Food, Hospitality & Travel",
        "Health & Wellness",
        "Industrial & Manufacturing",
        "Media & Entertainment",
        "Non-Profit & Government",
        "Other",
        "Retail & E-commerce"
    ],
    
    "page_type": [
        "Awareness & Discovery",
        "Consideration & Evaluation",
        "Conversion",
        "Internal & Navigation",
        "Post-Conversion & Other"
    ]
}


def validate_categories(business_model, customer_type, conversion_type, 
                       industry, page_type):
    """Validate that all categories are valid"""
    
    errors = []
    
    if business_model not in VALID_CATEGORIES["business_model"]:
        errors.append(f"Invalid business_model: {business_model}")
    
    if customer_type not in VALID_CATEGORIES["customer_type"]:
        errors.append(f"Invalid customer_type: {customer_type}")
    
    if conversion_type not in VALID_CATEGORIES["conversion_type"]:
        errors.append(f"Invalid conversion_type: {conversion_type}")
    
    if industry not in VALID_CATEGORIES["industry"]:
        errors.append(f"Invalid industry: {industry}")
    
    if page_type not in VALID_CATEGORIES["page_type"]:
        errors.append(f"Invalid page_type: {page_type}")
    
    if errors:
        raise ValueError("Category validation failed:\n" + "\n".join(errors))
    
    return True


# ============================================================================
# Error Handling Example
# ============================================================================

def safe_predict(control_image_path, variant_image_path,
                business_model, customer_type, conversion_type,
                industry, page_type):
    """
    Safe prediction with error handling and validation
    """
    
    try:
        # Validate categories first
        validate_categories(business_model, customer_type, conversion_type,
                          industry, page_type)
        
        # Make prediction
        result = predict_with_gradio_client(
            control_image_path=control_image_path,
            variant_image_path=variant_image_path,
            business_model=business_model,
            customer_type=customer_type,
            conversion_type=conversion_type,
            industry=industry,
            page_type=page_type
        )
        
        return {
            "success": True,
            "result": result
        }
        
    except ValueError as e:
        return {
            "success": False,
            "error": "Validation Error",
            "message": str(e)
        }
    
    except Exception as e:
        return {
            "success": False,
            "error": "API Error",
            "message": str(e)
        }