modules/proactive_learning.py

"""

Proactive Learning Assistance Module (Phase 1)

Implements intelligent prompting suggestions, context-aware follow-up questions, and critical knowledge gap identification

"""

import json
from typing import Dict, List, Optional, Tuple
from datetime import datetime
from openai import OpenAI


class KnowledgeGapAnalyzer:
    """Analyzes user knowledge gaps, especially critical safety-related gaps"""
    
    # Safety-critical ADAS features that require high knowledge levels
    SAFETY_CRITICAL_FEATURES = [
        "Function of Active Distance Assist DISTRONIC",
        "Function of Active Stop-and-Go Assist",
        "Function of Active Steering Assist"
    ]
    
    # Knowledge level thresholds
    CRITICAL_GAP_THRESHOLD = 0.5  # Below this is considered a critical gap for safety features
    WEAK_AREA_THRESHOLD = 0.6  # Below this is considered a weak area
    
    def __init__(self, available_topics: List[str]):
        self.available_topics = available_topics
    
    def identify_critical_gaps(self, user_profile) -> List[str]:
        """

        Identify critical knowledge gaps that could impact safety

        

        Returns:

            List of topics with critical knowledge gaps

        """
        critical_gaps = []
        knowledge_level = user_profile.knowledge_level if hasattr(user_profile, 'knowledge_level') else {}
        
        for topic in self.available_topics:
            level = knowledge_level.get(topic, 0.0)
            
            # Check if it's a safety-critical feature with low knowledge
            if topic in self.SAFETY_CRITICAL_FEATURES and level < self.CRITICAL_GAP_THRESHOLD:
                critical_gaps.append(topic)
        
        return critical_gaps
    
    def identify_weak_areas(self, user_profile) -> List[str]:
        """

        Identify all weak areas (not just critical)

        

        Returns:

            List of topics with weak knowledge levels

        """
        weak_areas = []
        knowledge_level = user_profile.knowledge_level if hasattr(user_profile, 'knowledge_level') else {}
        
        for topic in self.available_topics:
            level = knowledge_level.get(topic, 0.0)
            if level < self.WEAK_AREA_THRESHOLD:
                weak_areas.append(topic)
        
        return weak_areas
    
    def get_gap_priority(self, user_profile) -> List[Tuple[str, float]]:
        """

        Get knowledge gaps with priority scores

        

        Returns:

            List of (topic, priority_score) tuples, sorted by priority

        """
        gaps = []
        knowledge_level = user_profile.knowledge_level if hasattr(user_profile, 'knowledge_level') else {}
        
        for topic in self.available_topics:
            level = knowledge_level.get(topic, 0.0)
            
            # Calculate priority score
            priority = 0.0
            
            # Safety-critical features get higher priority
            if topic in self.SAFETY_CRITICAL_FEATURES:
                priority += 2.0
            
            # Lower knowledge level = higher priority
            priority += (1.0 - level) * 1.5
            
            # Check if it's in weak areas
            if hasattr(user_profile, 'weak_areas') and topic in user_profile.weak_areas:
                priority += 0.5
            
            gaps.append((topic, priority))
        
        # Sort by priority (descending)
        gaps.sort(key=lambda x: x[1], reverse=True)
        return gaps


class PromptSuggestionGenerator:
    """Generates intelligent prompt suggestions based on user profile and learning history"""
    
    def __init__(self, client: OpenAI, rag_engine, knowledge_gap_analyzer: KnowledgeGapAnalyzer, 

                 available_topics: List[str]):
        self.client = client
        self.rag_engine = rag_engine
        self.gap_analyzer = knowledge_gap_analyzer
        self.available_topics = available_topics
    
    def generate_suggestions(self, user_id: str, user_profile, learning_path=None, 

                            context: Optional[str] = None, max_suggestions: int = 5) -> List[Dict[str, str]]:
        """

        Generate prompt suggestions based on multiple criteria

        

        Args:

            user_id: User ID

            user_profile: UserProfile object

            learning_path: Optional LearningPath object

            context: Optional context (e.g., recent question)

            max_suggestions: Maximum number of suggestions to return

            

        Returns:

            List of suggestion dictionaries with 'question' and 'reason' keys

        """
        suggestions = []
        
        # 1. Based on critical knowledge gaps
        critical_gaps = self.gap_analyzer.identify_critical_gaps(user_profile)
        for topic in critical_gaps[:2]:  # Top 2 critical gaps
            question = self._generate_question_for_topic(topic, "beginner")
            if question:
                suggestions.append({
                    "question": question,
                    "reason": f"Critical Safety Feature: Your understanding of {topic.replace('Function of ', '')} needs improvement",
                    "priority": "high",
                    "type": "critical_gap"
                })
        
        # 2. Based on learning path
        if learning_path and hasattr(learning_path, 'nodes') and learning_path.nodes:
            current_node = None
            if learning_path.current_node_index < len(learning_path.nodes):
                current_node = learning_path.nodes[learning_path.current_node_index]
            
            if current_node and current_node.status != "completed":
                question = self._generate_question_for_topic(current_node.topic, current_node.bloom_level)
                if question:
                    suggestions.append({
                        "question": question,
                        "reason": f"Learning Path: Current learning node - {current_node.topic}",
                        "priority": "medium",
                        "type": "learning_path"
                    })
        
        # 3. Based on weak areas
        weak_areas = self.gap_analyzer.identify_weak_areas(user_profile)
        for topic in weak_areas[:2]:  # Top 2 weak areas
            if topic not in critical_gaps:  # Avoid duplicates
                question = self._generate_question_for_topic(topic, "understand")
                if question:
                    suggestions.append({
                        "question": question,
                        "reason": f"Weak Area: Recommend strengthening understanding of {topic.replace('Function of ', '')}",
                        "priority": "medium",
                        "type": "weak_area"
                    })
        
        # 4. Based on recent questions (if context provided)
        if context:
            related_questions = self._generate_related_questions(context)
            for q in related_questions[:2]:
                suggestions.append({
                    "question": q,
                    "reason": "Related Question: Explore deeper into the topic you just asked about",
                    "priority": "low",
                    "type": "related"
                })
        
        # 5. Based on unlearned topics
        knowledge_level = user_profile.knowledge_level if hasattr(user_profile, 'knowledge_level') else {}
        unlearned_topics = [t for t in self.available_topics if t not in knowledge_level]
        for topic in unlearned_topics[:1]:  # Top 1 unlearned topic
            question = self._generate_question_for_topic(topic, "remember")
            if question:
                suggestions.append({
                    "question": question,
                    "reason": f"New Topic: Start learning {topic.replace('Function of ', '')}",
                    "priority": "low",
                    "type": "new_topic"
                })
        
        # Rank and filter suggestions
        suggestions = self._rank_suggestions(suggestions)
        return suggestions[:max_suggestions]
    
    def _generate_question_for_topic(self, topic: str, level: str = "understand") -> Optional[str]:
        """Generate a question for a specific topic"""
        try:
            # Use RAG to get topic information
            query = f"What are the key points about {topic}?"
            answer, _ = self.rag_engine.query(query)
            
            # Generate question using LLM
            prompt = f"""Based on the following information about {topic}, generate a single, clear question that a user might ask to learn about this topic.



The question should be at a {level} level (from Bloom's taxonomy).



Information:

{answer[:500]}  # Limit context to avoid token limits



Generate only the question text, nothing else. The question should be:

- Clear and specific

- Appropriate for someone learning about ADAS systems

- In Chinese or English (match the user's language preference)



Question:"""
            
            response = self.client.chat.completions.create(
                model="gpt-4o-mini",
                messages=[
                    {"role": "system", "content": "You are a helpful assistant that generates educational questions."},
                    {"role": "user", "content": prompt}
                ],
                temperature=0.7,
                max_tokens=100
            )
            
            question = response.choices[0].message.content.strip()
            # Remove quotes if present
            question = question.strip('"').strip("'")
            return question
        except Exception as e:
            print(f"Error generating question for topic {topic}: {e}")
            # Fallback to simple question
            topic_clean = topic.replace("Function of ", "").replace(" Assist", "")
            return f"What is {topic_clean} and how does it work?"
    
    def _generate_related_questions(self, context: str) -> List[str]:
        """Generate related questions based on context"""
        try:
            prompt = f"""Based on the following question or context, generate 2-3 related follow-up questions that would help deepen understanding.



Context: {context[:300]}



Generate 2-3 questions, one per line. Questions should:

- Build upon the context

- Help explore related concepts

- Be clear and specific



Questions:"""
            
            response = self.client.chat.completions.create(
                model="gpt-4o-mini",
                messages=[
                    {"role": "system", "content": "You are a helpful assistant that generates educational follow-up questions."},
                    {"role": "user", "content": prompt}
                ],
                temperature=0.7,
                max_tokens=200
            )
            
            questions_text = response.choices[0].message.content.strip()
            questions = [q.strip().strip('-').strip() for q in questions_text.split('\n') if q.strip()]
            return questions[:3]
        except Exception as e:
            print(f"Error generating related questions: {e}")
            return []
    
    def _rank_suggestions(self, suggestions: List[Dict]) -> List[Dict]:
        """Rank suggestions by priority"""
        priority_weights = {"high": 3, "medium": 2, "low": 1}
        suggestions.sort(key=lambda x: priority_weights.get(x.get("priority", "low"), 1), reverse=True)
        return suggestions


class FollowUpQuestionGenerator:
    """Generates context-aware follow-up questions based on RAG answers"""
    
    def __init__(self, client: OpenAI, rag_engine):
        self.client = client
        self.rag_engine = rag_engine
        
        self.bloom_levels = ["remember", "understand", "apply", "analyze", "evaluate", "create"]
    
    def generate_follow_up_questions(self, answer: str, user_profile, 

                                     max_questions: int = 5) -> List[Dict[str, str]]:
        """

        Generate follow-up questions based on the answer provided

        

        Args:

            answer: The RAG answer text

            user_profile: UserProfile object

            max_questions: Maximum number of questions to generate

            

        Returns:

            List of question dictionaries with 'question' and 'bloom_level' keys

        """
        questions = []
        
        # Determine user's current Bloom level (default to understand)
        current_bloom = self._infer_user_bloom_level(user_profile)
        current_index = self.bloom_levels.index(current_bloom) if current_bloom in self.bloom_levels else 1
        
        # Generate questions for next 2-3 Bloom levels
        target_levels = self.bloom_levels[current_index:current_index + 3]
        
        for level in target_levels[:2]:  # Limit to 2 levels
            level_questions = self._generate_questions_by_bloom(answer, level)
            questions.extend(level_questions[:2])  # 2 questions per level
        
        # Also generate related concept questions
        related_questions = self._generate_related_concept_questions(answer)
        questions.extend(related_questions[:1])
        
        return questions[:max_questions]
    
    def _infer_user_bloom_level(self, user_profile) -> str:
        """Infer user's current Bloom taxonomy level based on profile"""
        # Check recent test performance
        if hasattr(user_profile, 'bloom_level_performance') and user_profile.bloom_level_performance:
            # Find the highest level where user has good performance
            for level in reversed(self.bloom_levels):
                for topic_perf in user_profile.bloom_level_performance.values():
                    if level in topic_perf and topic_perf[level] >= 0.7:
                        return level
        
        # Default based on overall progress
        if hasattr(user_profile, 'knowledge_level') and user_profile.knowledge_level:
            avg_level = sum(user_profile.knowledge_level.values()) / len(user_profile.knowledge_level.values())
            if avg_level < 0.3:
                return "remember"
            elif avg_level < 0.6:
                return "understand"
            else:
                return "apply"
        
        return "understand"  # Default
    
    def _generate_questions_by_bloom(self, answer: str, bloom_level: str) -> List[Dict[str, str]]:
        """Generate questions at a specific Bloom taxonomy level"""
        try:
            bloom_descriptions = {
                "remember": "test basic recall of facts and information",
                "understand": "test explanation and interpretation of concepts",
                "apply": "test application of knowledge in practical situations",
                "analyze": "test analysis of relationships and structure",
                "evaluate": "test evaluation and judgment based on criteria",
                "create": "test creation of new ideas or solutions"
            }
            
            prompt = f"""Based on the following answer about ADAS systems, generate 2 follow-up questions at the {bloom_level} level of Bloom's taxonomy.



Bloom level description: {bloom_descriptions.get(bloom_level, '')}



Answer text:

{answer[:800]}  # Limit context



Generate 2 questions that:

- Build upon the information in the answer

- Are at the {bloom_level} level

- Help deepen understanding

- Are clear and specific



Output format: One question per line, no numbering or bullets.



Questions:"""
            
            response = self.client.chat.completions.create(
                model="gpt-4o-mini",
                messages=[
                    {"role": "system", "content": "You are an educational assistant that generates follow-up questions."},
                    {"role": "user", "content": prompt}
                ],
                temperature=0.7,
                max_tokens=200
            )
            
            questions_text = response.choices[0].message.content.strip()
            question_list = [q.strip().strip('-').strip() for q in questions_text.split('\n') if q.strip()]
            
            return [{"question": q, "bloom_level": bloom_level} for q in question_list[:2]]
        except Exception as e:
            print(f"Error generating questions by Bloom level: {e}")
            return []
    
    def _generate_related_concept_questions(self, answer: str) -> List[Dict[str, str]]:
        """Generate questions about related concepts"""
        try:
            prompt = f"""Based on the following answer, generate 1 question about a related ADAS concept that would help the user understand the broader context.



Answer:

{answer[:500]}



Generate 1 question about a related concept or feature that connects to the information provided.



Question:"""
            
            response = self.client.chat.completions.create(
                model="gpt-4o-mini",
                messages=[
                    {"role": "system", "content": "You are an educational assistant."},
                    {"role": "user", "content": prompt}
                ],
                temperature=0.7,
                max_tokens=100
            )
            
            question = response.choices[0].message.content.strip().strip('"').strip("'")
            return [{"question": question, "bloom_level": "understand"}]
        except Exception as e:
            print(f"Error generating related concept question: {e}")
            return []


class ProactiveLearningEngine:
    """Main engine for proactive learning assistance"""
    
    def __init__(self, client: OpenAI, rag_engine, user_profiling, adaptive_engine=None, 

                 available_topics: List[str] = None):
        self.client = client
        self.rag_engine = rag_engine
        self.user_profiling = user_profiling
        self.adaptive_engine = adaptive_engine
        self.available_topics = available_topics or []
        
        # Initialize components
        self.gap_analyzer = KnowledgeGapAnalyzer(self.available_topics)
        self.suggestion_generator = PromptSuggestionGenerator(
            client, rag_engine, self.gap_analyzer, self.available_topics
        )
        self.followup_generator = FollowUpQuestionGenerator(client, rag_engine)
    
    def get_prompt_suggestions(self, user_id: str, context: Optional[str] = None, 

                              max_suggestions: int = 5) -> List[Dict[str, str]]:
        """

        Get prompt suggestions for a user

        

        Args:

            user_id: User ID

            context: Optional context (e.g., recent question)

            max_suggestions: Maximum number of suggestions

            

        Returns:

            List of suggestion dictionaries

        """
        if not self.user_profiling:
            return []
        
        user_profile = self.user_profiling.get_or_create_profile(user_id)
        
        # Get learning path if available
        learning_path = None
        if self.adaptive_engine:
            learning_path = self.adaptive_engine.get_active_path(user_id)
        
        return self.suggestion_generator.generate_suggestions(
            user_id, user_profile, learning_path, context, max_suggestions
        )
    
    def get_follow_up_questions(self, user_id: str, answer: str, 

                                max_questions: int = 5) -> List[Dict[str, str]]:
        """

        Get follow-up questions based on an answer

        

        Args:

            user_id: User ID

            answer: The RAG answer text

            max_questions: Maximum number of questions

            

        Returns:

            List of question dictionaries

        """
        if not self.user_profiling:
            return []
        
        user_profile = self.user_profiling.get_or_create_profile(user_id)
        return self.followup_generator.generate_follow_up_questions(
            answer, user_profile, max_questions
        )
    
    def get_critical_gaps(self, user_id: str) -> List[str]:
        """

        Get critical knowledge gaps for a user

        

        Args:

            user_id: User ID

            

        Returns:

            List of topics with critical gaps

        """
        if not self.user_profiling:
            return []
        
        user_profile = self.user_profiling.get_or_create_profile(user_id)
        return self.gap_analyzer.identify_critical_gaps(user_profile)
    
    def analyze_user_state(self, user_id: str) -> Dict:
        """

        Analyze user's current learning state

        

        Args:

            user_id: User ID

            

        Returns:

            Dictionary with analysis results

        """
        if not self.user_profiling:
            return {}
        
        user_profile = self.user_profiling.get_or_create_profile(user_id)
        
        critical_gaps = self.gap_analyzer.identify_critical_gaps(user_profile)
        weak_areas = self.gap_analyzer.identify_weak_areas(user_profile)
        gap_priorities = self.gap_analyzer.get_gap_priority(user_profile)
        
        return {
            "critical_gaps": critical_gaps,
            "weak_areas": weak_areas,
            "gap_priorities": gap_priorities[:5],  # Top 5
            "total_gaps": len(weak_areas),
            "critical_gaps_count": len(critical_gaps)
        }