"""Platform/CMS/Technology detector for websites."""

import re
from typing import Dict, Optional, List
from urllib.parse import urlparse


class PlatformDetector:
    """Detect platform, CMS, and technology stack of a website."""
    
    # Platform detection patterns
    PLATFORM_PATTERNS = {
        'shopify': {
            'indicators': [
                r'shopify\.com',
                r'cdn\.shopify\.com',
                r'\.myshopify\.com',
                r'shopify-analytics',
                r'shopify\.theme',
                r'Shopify\.checkout',
            ],
            'name': 'Shopify',
            'type': 'ecommerce_platform',
            'cms': 'Shopify',
            'tech': ['Liquid', 'React', 'Vue.js'],
            'features': {
                'collections': True,
                'categories': False,
                'api_available': True,
                'requires_js': True
            }
        },
        'woocommerce': {
            'indicators': [
                r'woocommerce',
                r'wp-content/plugins/woocommerce',
                r'wc-',
                r'add-to-cart\.php',
                r'woocommerce-active',
            ],
            'name': 'WooCommerce',
            'type': 'ecommerce_platform',
            'cms': 'WordPress',
            'tech': ['WordPress', 'PHP', 'MySQL'],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': True,
                'requires_js': False
            }
        },
        'wordpress': {
            'indicators': [
                r'wp-content',
                r'wp-includes',
                r'wordpress',
                r'/wp-admin/',
                r'wp-json',
            ],
            'name': 'WordPress',
            'type': 'cms',
            'cms': 'WordPress',
            'tech': ['WordPress', 'PHP', 'MySQL'],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': True,
                'requires_js': False
            }
        },
        'magento': {
            'indicators': [
                r'magento',
                r'/static/version',
                r'requirejs-config\.js',
                r'Magento_',
            ],
            'name': 'Magento',
            'type': 'ecommerce_platform',
            'cms': 'Magento',
            'tech': ['PHP', 'MySQL', 'RequireJS'],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': True,
                'requires_js': True
            }
        },
        'bigcommerce': {
            'indicators': [
                r'bigcommerce\.com',
                r'cdn\.bigcommerce\.com',
                r'stencil\.bigcommerce\.com',
            ],
            'name': 'BigCommerce',
            'type': 'ecommerce_platform',
            'cms': 'BigCommerce',
            'tech': ['Stencil', 'Handlebars'],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': True,
                'requires_js': True
            }
        },
        'prestashop': {
            'indicators': [
                r'prestashop',
                r'/modules/',
                r'presta',
            ],
            'name': 'PrestaShop',
            'type': 'ecommerce_platform',
            'cms': 'PrestaShop',
            'tech': ['PHP', 'MySQL', 'Smarty'],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': True,
                'requires_js': False
            }
        },
        'redbubble': {
            'indicators': [
                r'redbubble\.com',
                r'redbubble',
                r'/shop/',
                r'/products/',
            ],
            'name': 'Redbubble',
            'type': 'marketplace',
            'cms': 'Redbubble',
            'tech': ['React', 'JavaScript', 'Node.js'],
            'features': {
                'collections': True,
                'categories': True,
                'api_available': False,
                'requires_js': True
            }
        },
    }
    
    def detect(self, html_content: str, url: str, headers: Optional[Dict] = None) -> Dict:
        """
        Detect platform from HTML content, URL, and headers.
        
        Args:
            html_content: HTML content of the page
            url: Website URL
            headers: HTTP headers (optional)
            
        Returns:
            Dictionary with platform information
        """
        html_lower = html_content.lower()
        url_lower = url.lower()
        
        # Check each platform
        for platform_key, platform_info in self.PLATFORM_PATTERNS.items():
            for indicator in platform_info['indicators']:
                if re.search(indicator, html_lower, re.IGNORECASE) or re.search(indicator, url_lower, re.IGNORECASE):
                    return {
                        'platform': platform_info['name'],
                        'platform_key': platform_key,
                        'type': platform_info['type'],
                        'cms': platform_info['cms'],
                        'tech_stack': platform_info['tech'],
                        'features': platform_info['features'],
                        'confidence': 'high'
                    }
        
        # Check for generic indicators
        detected = {
            'platform': 'Unknown',
            'platform_key': 'unknown',
            'type': 'generic',
            'cms': 'Unknown',
            'tech_stack': [],
            'features': {
                'collections': False,
                'categories': True,
                'api_available': False,
                'requires_js': False
            },
            'confidence': 'low'
        }
        
        # Detect JavaScript frameworks
        if re.search(r'react|react-dom', html_lower):
            detected['tech_stack'].append('React')
        if re.search(r'vue\.|vuejs', html_lower):
            detected['tech_stack'].append('Vue.js')
        if re.search(r'angular', html_lower):
            detected['tech_stack'].append('Angular')
        
        # Detect if requires JS
        if re.search(r'__next__|next\.js', html_lower):
            detected['tech_stack'].append('Next.js')
            detected['features']['requires_js'] = True
        
        return detected