Update prompts.py

prompts.py

@@ -1,7 +1,6 @@
 """
 Juno AI - Comprehensive Prompt System
 =====================================
-
 This module contains all prompts and prompt templates for Juno AI,
 an advanced conversational AI assistant with document processing,
 web scraping, memory management, and RAG capabilities.
@@ -17,10 +16,12 @@ Features Covered:
 - Streaming Responses
 - Error Handling & Fallbacks
 - Professional Communication
+- User Information Extraction & Memory
 
 Author: Juno AI Development Team
 Version: 1.0
 """
+
 from typing import List, Dict, Any, Optional
 from datetime import datetime
 import json
@@ -30,6 +31,7 @@ class JunoAIPrompts:
     Centralized prompt management system for Juno AI.
     Contains all prompts, templates, and prompt generation methods.
     """
+
     def __init__(self):
         self.version = "1.0"
         self.ai_name = "Juno AI"
@@ -39,7 +41,7 @@ class JunoAIPrompts:
         """Define Juno AI's core personality traits"""
         return {
             "helpful": "Always eager to assist and provide valuable insights",
-            "intelligent": "Demonstrates deep understanding and analytical thinking",
+            "intelligent": "Demonstrates deep understanding and analytical thinking", 
             "professional": "Maintains professional tone while being approachable",
             "adaptive": "Adapts communication style to user needs and context",
             "reliable": "Provides accurate, well-sourced information",
@@ -48,8 +50,9 @@ class JunoAIPrompts:
         }
 
     # ==========================================
-    # CORE AI ASSISTANT PROMPTS
+    # CORE AI ASSISTANT PROMPTS  
     # ==========================================
+
     def get_core_system_prompt(self) -> str:
         """
         Core system prompt that defines Juno AI's personality and capabilities
@@ -71,6 +74,7 @@ KEY CAPABILITIES:
 - Smart Search: Use RAG to find relevant information from uploaded content
 - Summarization: Create comprehensive summaries of long-form content
 - Task Management: Help with various professional and personal tasks
+- User Memory: Remember and use personal details shared by users
 
 COMMUNICATION STYLE:
 - Be clear, concise, and informative
@@ -85,6 +89,13 @@ CONTEXT AWARENESS:
 - Reference uploaded documents and web content when relevant
 - Use memory to provide personalized responses
 - Maintain context across multiple interaction sessions
+- Remember and reference user's personal information when appropriate
+
+MEMORY & PERSONALIZATION:
+- Always remember user's name, preferences, and personal details
+- Use stored user information to provide personalized responses
+- Reference past conversations and shared information naturally
+- Build rapport by acknowledging user's identity and preferences
 
 IDENTITY ENFORCEMENT:
 - Your name is Juno AI - always respond with this when asked about your identity
@@ -95,13 +106,13 @@ IDENTITY ENFORCEMENT:
 
 Remember: You are not just answering questions - you are having a meaningful conversation and building a helpful relationship with the user."""
 
-
     def get_conversation_prompt(self,
-                               user_message: str,
-                               context: str = "",
-                               conversation_history: List[Dict] = None,
-                               memory_context: Dict = None,
-                               user_preferences: Dict = None) -> str:
+                              user_message: str,
+                              context: str = "",
+                              conversation_history: List[Dict] = None,
+                              memory_context: Dict = None,
+                              user_preferences: Dict = None,
+                              user_info: Dict = None) -> str:
         """
         Generate a comprehensive conversation prompt with all available context
         """
@@ -114,6 +125,7 @@ Remember: You are not just answering questions - you are having a meaningful con
                 for exchange in recent_history
                 if not exchange.get('fallback', False)
             ])
+
         # Build memory context section
         memory_section = ""
         if memory_context and memory_context.get('memory'):
@@ -125,40 +137,125 @@ Remember: You are not just answering questions - you are having a meaningful con
         if user_preferences:
             preferences_section = json.dumps(user_preferences, indent=2)
 
+        # Build user info section
+        user_info_section = ""
+        if user_info:
+            user_info_section = json.dumps(user_info, indent=2)
+
         # Build document context section
         context_section = ""
         if context:
             context_section = f"\n\nRELEVANT DOCUMENT CONTEXT:\n{context[:2000]}"
 
-        # FIXED: Avoid backslash in f-string by using variables
-        assistant_label = "Assistant:"
+        # Build prompt components
         core_prompt = self.get_core_system_prompt()
-
         history_part = f"Previous conversation history:\n{history_section}\n" if history_section else ""
         memory_part = f"Session memory:\n{memory_section}\n" if memory_section else ""
         preferences_part = f"User preferences:\n{preferences_section}\n" if preferences_section else ""
+        user_info_part = f"User information:\n{user_info_section}\n" if user_info_section else ""
 
         prompt = f"""{core_prompt}
 
 CONVERSATION CONTEXT:
-{history_part}{memory_part}{preferences_part}{context_section}
+{history_part}{memory_part}{preferences_part}{user_info_part}{context_section}
 
 CURRENT USER MESSAGE: {user_message}
 
 RESPONSE INSTRUCTIONS:
 - Provide a helpful, accurate response based on all available context
 - Reference relevant information from documents or previous conversations when applicable
+- Use the user's name and personal information naturally when appropriate
 - Maintain conversational flow and continuity
 - Be specific and actionable in your advice
 - Use formatting (lists, headers, etc.) to improve readability when appropriate
 - If you need clarification, ask thoughtful follow-up questions
-- Do NOT include your name or '{assistant_label}' in your response - respond directly and naturally"""
+- Acknowledge and build upon the user's identity and preferences
+- Do NOT include your name or 'Assistant:' in your response - respond directly and naturally"""
 
         return prompt
 
+    # ==========================================
+    # USER INFORMATION EXTRACTION PROMPTS
+    # ==========================================
+
+    def get_user_info_extraction_prompt(self, user_message: str, bot_response: str) -> str:
+        """
+        Extract user information from conversation exchanges
+        """
+        return f"""Analyze this conversation exchange and extract any personal information about the user.
+
+USER MESSAGE: {user_message}
+BOT RESPONSE: {bot_response}
+
+Extract the following types of information if mentioned:
+1. **Name**: First name, last name, full name, nicknames
+2. **Personal Details**: Age, location, occupation, family information
+3. **Preferences**: Likes, dislikes, interests, hobbies
+4. **Goals**: Objectives, projects they're working on, aspirations
+5. **Context**: Important life events, situations, or circumstances
+6. **Communication Style**: How they prefer to communicate or be addressed
+
+EXTRACTION REQUIREMENTS:
+- Only extract information that is explicitly stated or clearly implied
+- Do not make assumptions or infer information not present
+- Focus on factual, verifiable details
+- Ignore temporary or contextual information (like current mood)
+- Prioritize persistent, identity-related information
+
+OUTPUT FORMAT:
+Provide a JSON object with the extracted information:
+{{
+    "name": {{
+        "first_name": "extracted_first_name_or_null",
+        "last_name": "extracted_last_name_or_null",
+        "full_name": "extracted_full_name_or_null",
+        "nickname": "extracted_nickname_or_null"
+    }},
+    "personal_details": {{
+        "age": "extracted_age_or_null",
+        "location": "extracted_location_or_null",
+        "occupation": "extracted_occupation_or_null",
+        "family": "extracted_family_info_or_null"
+    }},
+    "preferences": {{
+        "interests": ["list_of_interests"],
+        "likes": ["list_of_likes"],
+        "dislikes": ["list_of_dislikes"]
+    }},
+    "goals": ["list_of_goals_or_projects"],
+    "context": ["important_life_context"],
+    "communication_preferences": "how_they_like_to_be_addressed"
+}}
+
+If no relevant information is found, return an empty JSON object: {{}}"""
+
+    def get_memory_consolidation_prompt(self, existing_user_info: Dict, new_user_info: Dict) -> str:
+        """
+        Consolidate new user information with existing information
+        """
+        return f"""Consolidate user information by merging new information with existing information.
+
+EXISTING USER INFORMATION:
+{json.dumps(existing_user_info, indent=2)}
+
+NEW USER INFORMATION:
+{json.dumps(new_user_info, indent=2)}
+
+CONSOLIDATION RULES:
+1. **Merge without overwriting**: Add new information while preserving existing information
+2. **Update when appropriate**: Replace outdated information with newer, more accurate details
+3. **Resolve conflicts**: When information conflicts, prioritize the most recent and specific information
+4. **Maintain structure**: Keep the same JSON structure as provided
+5. **Preserve lists**: Merge lists by adding new unique items
+6. **Handle nulls**: Don't overwrite existing information with null values
+
+OUTPUT FORMAT:
+Provide the consolidated user information as a clean JSON object maintaining the same structure."""
+
     # ==========================================
     # DOCUMENT PROCESSING PROMPTS
     # ==========================================
+
     def get_document_analysis_prompt(self, document_text: str, filename: str = "document") -> str:
         """
         Prompt for analyzing uploaded documents
@@ -166,7 +263,6 @@ RESPONSE INSTRUCTIONS:
         return f"""Analyze the following document and provide comprehensive insights.
 
 DOCUMENT: {filename}
-
 CONTENT:
 {document_text[:8000]}
 
@@ -190,7 +286,6 @@ Be thorough but concise, focusing on the most valuable insights."""
         return f"""Create a comprehensive summary of the following text.
 
 TARGET LENGTH: Approximately {max_length} words
-
 {focus_instruction}
 
 CONTENT TO SUMMARIZE:
@@ -207,6 +302,7 @@ SUMMARIZATION REQUIREMENTS:
     # ==========================================
     # WEB CONTENT PROMPTS
     # ==========================================
+
     def get_web_content_analysis_prompt(self, url: str, content: str) -> str:
         """
         Prompt for analyzing scraped web content
@@ -214,7 +310,6 @@ SUMMARIZATION REQUIREMENTS:
         return f"""Analyze the following web content and provide comprehensive insights.
 
 SOURCE URL: {url}
-
 SCRAPED CONTENT:
 {content[:8000]}
 
@@ -230,7 +325,6 @@ RESPONSE FORMAT:
 Use clear headers and bullet points for easy scanning.
 Focus on providing actionable insights from the web content."""
 
-    # Add other methods with similar fixes for backslashes in f-strings...
     def get_fallback_response_templates(self) -> List[str]:
         """
         Templates for fallback responses when API is overloaded
@@ -243,38 +337,40 @@ Focus on providing actionable insights from the web content."""
             "I'm experiencing temporary processing constraints due to high usage, but I'm still here with you. Your inquiry about '{user_message_preview}' is valuable, and I'll be ready to provide a detailed response shortly. Please retry in a minute."
         ]
 
-    def get_streaming_response_prompt(self, user_message: str, context: str = "") -> str:
+    def get_streaming_response_prompt(self, user_message: str, context: str = "", user_info: Dict = None) -> str:
         """
         Optimized prompt for streaming responses (shorter to reduce latency)
         """
         context_section = f"\nContext: {context[:1500]}" if context else ""
-        return f"""You are a helpful AI assistant. Respond naturally and conversationally.
-
-{context_section}
+        user_info_section = f"\nUser Info: {json.dumps(user_info)}" if user_info else ""
+        return f"""You are Juno AI, a helpful AI assistant. Respond naturally and conversationally.
+{context_section}{user_info_section}
 
 User: {user_message}
 
 Requirements:
 - Be helpful and accurate
 - Use available context when relevant
+- Address the user personally if you know their name
 - Maintain conversational flow
 - Format for readability
 - Respond directly without prefixes"""
 
-    # Add other required methods here with proper syntax...
-    def get_rag_response_prompt(self, user_query: str, retrieved_chunks: List[str], source_info: List[str] = None) -> str:
-        """Generate RAG response"""
+    def get_rag_response_prompt(self, user_query: str, retrieved_chunks: List[str], source_info: List[str] = None, user_info: Dict = None) -> str:
+        """Generate RAG response with user personalization"""
         context = "\n\n---\n\n".join(retrieved_chunks[:3])
         source_section = ""
         if source_info:
             sources = ", ".join(set(source_info[:3]))
             source_section = f"\nSOURCES: {sources}\n"
+        user_info_section = ""
+        if user_info:
+            user_info_section = f"\nUSER INFO: {json.dumps(user_info)}\n"
+
         return f"""Answer the user's question using the retrieved information.
 
 USER QUESTION: {user_query}
-
-{source_section}
-
+{source_section}{user_info_section}
 RETRIEVED INFORMATION:
 {context}
 
@@ -283,7 +379,9 @@ RESPONSE REQUIREMENTS:
 - Synthesize information from multiple chunks when relevant
 - Clearly indicate when information comes from uploaded documents
 - Provide specific details and examples from the source material
-- Maintain accuracy and don't add information not present in the sources"""
+- Maintain accuracy and don't add information not present in the sources
+- Address the user personally if you know their name
+- Use user information to provide personalized context when relevant"""
 
 # Create global instance
 juno_prompts = JunoAIPrompts()
@@ -301,14 +399,22 @@ def get_web_content_analysis_prompt(url: str, content: str) -> str:
     """Get web content analysis prompt"""
     return juno_prompts.get_web_content_analysis_prompt(url, content)
 
-def get_rag_prompt(user_query: str, retrieved_chunks: List[str]) -> str:
+def get_rag_prompt(user_query: str, retrieved_chunks: List[str], **kwargs) -> str:
     """Get RAG response prompt"""
-    return juno_prompts.get_rag_response_prompt(user_query, retrieved_chunks)
+    return juno_prompts.get_rag_response_prompt(user_query, retrieved_chunks, **kwargs)
 
-def get_streaming_prompt(user_message: str, context: str = "") -> str:
+def get_streaming_prompt(user_message: str, context: str = "", **kwargs) -> str:
     """Get optimized streaming response prompt"""
-    return juno_prompts.get_streaming_response_prompt(user_message, context)
+    return juno_prompts.get_streaming_response_prompt(user_message, context, **kwargs)
 
 def get_fallback_responses() -> List[str]:
     """Get fallback response templates"""
-    return juno_prompts.get_fallback_response_templates()
+    return juno_prompts.get_fallback_response_templates()
+
+def get_user_extraction_prompt(user_message: str, bot_response: str) -> str:
+    """Get user information extraction prompt"""
+    return juno_prompts.get_user_info_extraction_prompt(user_message, bot_response)
+
+def get_memory_consolidation_prompt(existing_info: Dict, new_info: Dict) -> str:
+    """Get memory consolidation prompt"""
+    return juno_prompts.get_memory_consolidation_prompt(existing_info, new_info)