core/validator.py

"""
Input Validation and Security Module
Handles prompt injection detection and input validation
"""

import re
from typing import Tuple
from config import HARMFUL_KEYWORDS, MAX_INPUT_LENGTH


class InputValidator:
    """Validates user inputs for security and quality."""
    
    @staticmethod
    def validate_input(text: str) -> Tuple[bool, str]:
        """
        Validate user input for security and quality.
        
        Args:
            text: Input text to validate
        
        Returns:
            Tuple of (is_valid, error_message)
        """
        # Check if empty
        if not text or not text.strip():
            return False, "Input cannot be empty"
        
        # Check length
        if len(text) > MAX_INPUT_LENGTH:
            return False, f"Input too long (max {MAX_INPUT_LENGTH} characters)"
        
        # Check for prompt injection
        is_safe, msg = InputValidator.detect_prompt_injection(text)
        if not is_safe:
            return False, msg
        
        # Check for minimum meaningful content
        if len(text.strip().split()) < 3:
            return False, "Input too short. Please provide more context"
        
        return True, "Valid input"
    
    @staticmethod
    def detect_prompt_injection(text: str) -> Tuple[bool, str]:
        """
        Detect potential prompt injection attacks.
        
        Uses blacklist of harmful keywords and patterns.
        
        Args:
            text: Text to check for prompt injection
        
        Returns:
            Tuple of (is_safe, warning_message)
        """
        text_lower = text.lower()
        
        # Check for harmful keywords
        for keyword in HARMFUL_KEYWORDS:
            if keyword.lower() in text_lower:
                return False, f"⚠️ Blocked: Detected suspicious pattern: '{keyword}'"
        
        # Check for common injection patterns
        injection_patterns = [
            r'```[\s\S]*?```',  # Code blocks (potential instruction override)
            r'<!--[\s\S]*?-->',  # HTML comments
            r'\[SYSTEM\]',  # System tokens
            r'\[IGNORE\]',  # Ignore directives
            r'\\x[0-9a-f]{2}',  # Hex encoding attempts
        ]
        
        for pattern in injection_patterns:
            if re.search(pattern, text, re.IGNORECASE):
                return False, "⚠️ Blocked: Detected suspicious instruction pattern"
        
        return True, "Safe input"
    
    @staticmethod
    def sanitize_input(text: str) -> str:
        """
        Sanitize input by removing potentially harmful characters.
        
        Args:
            text: Text to sanitize
        
        Returns:
            Sanitized text
        """
        # Remove null bytes
        text = text.replace('\x00', '')
        
        # Remove control characters except newlines and tabs
        text = ''.join(
            char for char in text 
            if ord(char) >= 32 or char in '\n\t'
        )
        
        # Remove multiple consecutive newlines
        text = re.sub(r'\n\n+', '\n\n', text)
        
        return text.strip()
    
    @staticmethod
    def validate_file_path(file_path: str) -> Tuple[bool, str]:
        """
        Validate file path for security.
        
        Args:
            file_path: File path to validate
        
        Returns:
            Tuple of (is_valid, error_message)
        """
        if not file_path:
            return False, "File path cannot be empty"
        
        # Check for path traversal attempts
        if ".." in file_path:
            return False, "Invalid file path: path traversal detected"
        
        # Check if path is absolute and outside project
        import os
        if os.path.isabs(file_path):
            project_root = os.path.dirname(os.path.abspath(__file__))
            if not os.path.abspath(file_path).startswith(project_root):
                return False, "Invalid file path: outside project directory"
        
        return True, "Valid file path"


class ContentValidator:
    """Validates content quality and relevance."""
    
    @staticmethod
    def is_meaningful_response(text: str, min_words: int = 10) -> bool:
        """
        Check if response is meaningful.
        
        Args:
            text: Text to validate
            min_words: Minimum words required
        
        Returns:
            True if meaningful, False otherwise
        """
        words = text.strip().split()
        return len(words) >= min_words
    
    @staticmethod
    def estimate_quality(text: str) -> float:
        """
        Estimate quality of a response (0.0 to 1.0).
        
        Args:
            text: Text to assess
        
        Returns:
            Quality score
        """
        score = 0.0
        
        # Length factor (max 0.3)
        word_count = len(text.split())
        length_score = min(word_count / 100, 1.0) * 0.3
        score += length_score
        
        # Structure factor (0.3) - presence of punctuation
        punctuation_count = sum(1 for c in text if c in '.!?;:')
        structure_score = min(punctuation_count / 10, 1.0) * 0.3
        score += structure_score
        
        # Diversity factor (0.4) - lexical diversity
        words = text.lower().split()
        unique_words = len(set(words))
        if len(words) > 0:
            diversity_score = (unique_words / len(words)) * 0.4
            score += diversity_score
        
        return min(score, 1.0)

    @staticmethod
    def is_acceptable_summary(text: str, min_chars: int = 80) -> bool:
        """
        Validate summary quality with flexible structured-output support.

        Args:
            text: Summary text
            min_chars: Minimum character threshold

        Returns:
            True when summary is acceptable
        """
        if not text:
            return False

        cleaned = text.strip()
        if len(cleaned) >= min_chars:
            return True

        lower = cleaned.lower()
        required_sections = [
            "key points",
            "main concept",
            "important details",
            "conclusion",
        ]
        has_all_sections = all(section in lower for section in required_sections)

        # Structured summaries can be concise but still useful.
        if has_all_sections and len(cleaned.split()) >= 20:
            return True

        return False


class PromptValidator:
    """Validates and refines prompts for consistency."""
    
    @staticmethod
    def validate_prompt_mode(mode: str) -> Tuple[bool, str]:
        """
        Validate if prompt mode is supported.
        
        Args:
            mode: Mode name
        
        Returns:
            Tuple of (is_valid, message)
        """
        valid_modes = ["normal", "detailed", "teacher", "exam"]
        
        if mode.lower() in valid_modes:
            return True, f"Valid mode: {mode}"
        else:
            return False, f"Invalid mode: {mode}. Choose from: {', '.join(valid_modes)}"
    
    @staticmethod
    def validate_feature(feature: str) -> Tuple[bool, str]:
        """
        Validate if feature is supported.
        
        Args:
            feature: Feature name
        
        Returns:
            Tuple of (is_valid, message)
        """
        valid_features = ["summarizer", "quiz", "explainer", "doubt_solver"]
        
        if feature.lower() in valid_features:
            return True, f"Valid feature: {feature}"
        else:
            return False, f"Invalid feature: {feature}. Choose from: {', '.join(valid_features)}"