utils/config.py

import os
import re
import urllib.parse
from dotenv import load_dotenv

class Config:
    def __init__(self):
        load_dotenv()
        
        # Detect if running on HF Spaces
        self.is_hf_space = bool(os.getenv("SPACE_ID"))
        
        # API tokens
        self.hf_token = os.getenv("HF_TOKEN")
        self.openai_api_key = os.getenv("OPENAI_API_KEY")
        
        # API endpoints
        self.hf_api_url = self._sanitize_url(os.getenv("HF_API_ENDPOINT_URL", "https://api-inference.huggingface.co/v1/"))
        
        # Fallback settings
        self.use_fallback = os.getenv("USE_FALLBACK", "true").lower() == "true"
        
        # Local model configuration
        self.local_model_name = os.getenv("LOCAL_MODEL_NAME", "mistral:latest")
        self.ollama_host = self._sanitize_url(os.getenv("OLLAMA_HOST", ""))
        
        # OpenWeather API
        self.openweather_api_key = os.getenv("OPENWEATHER_API_KEY")
        
    def _sanitize_url(self, url: str) -> str:
        """Sanitize URL by removing whitespace and control characters"""
        if not url:
            return ""
            
        # Remove leading/trailing whitespace and control characters
        url = url.strip()
        # Remove any newlines, carriage returns, tabs, and null bytes
        url = re.sub(r'[\r\n\t\0\x0b\x0c]+', '', url)
        # Remove any trailing URL encoding artifacts
        url = url.rstrip('%0a').rstrip('%0d')
        
        # Handle the specific case where environment variable name is included
        if 'hf_api_endpoint_url=' in url.lower():
            # Extract the actual URL part
            url = url.split('=')[-1]
        
        # Remove any protocol prefix issues
        url = url.replace('hf_api_endpoint_url=', '')
        url = url.replace('HF_API_ENDPOINT_URL=', '')
        
        # Ensure proper URL format
        if url and not url.startswith(('http://', 'https://')):
            # Check if it looks like a domain
            if re.match(r'^[a-zA-Z0-9.-]+(?:\.[a-zA-Z]{2,})+', url) or 'huggingface.cloud' in url:
                url = 'https://' + url
            else:
                url = 'https://' + url
        
        # Remove double slashes (but keep ://)
        if '://' in url:
            protocol, rest = url.split('://', 1)
            rest = re.sub(r'//+', '/', rest)
            url = protocol + '://' + rest
            
        return url
        
    def _sanitize_host(self, host: str) -> str:
        """Sanitize host by removing whitespace and control characters"""
        if not host:
            return ""
            
        # Remove leading/trailing whitespace and control characters
        host = host.strip()
        # Remove any newlines, carriage returns, tabs, and null bytes
        host = re.sub(r'[\r\n\t\0\x0b\x0c]+', '', host)
        # Remove any trailing URL encoding artifacts
        host = host.rstrip('%0a').rstrip('%0d')
        
        return host

# Global config instance
config = Config()