agents/chat_agent.py

#!/usr/bin/env python3
"""
Flexible LangGraph agent for cyber-legal assistant
Agent can call tools, process results, and decide to continue or answer
"""

import os
import copy
import logging
import traceback
from typing import Dict, Any, List, Optional
from datetime import datetime
from langgraph.graph import StateGraph, END
from langchain_core.messages import HumanMessage, SystemMessage, AIMessage, ToolMessage

logger = logging.getLogger(__name__)
from agent_states.agent_state import AgentState
from utils.utils_fn import PerformanceMonitor, find_tool
from utils.tools import tools, tools_for_client, tools_for_lawyer


class HexiagonAgent:
    def __init__(self, llm, tools: List[Any] = tools, tools_facade: List[Any] = tools):
        self.tools = tools
        self.tools_facade = tools_facade
        self.llm = llm
        self.performance_monitor = PerformanceMonitor()
        # Constrain to one tool call at a time
        self.llm_with_tools = self.llm.bind_tools(self.tools_facade,)
        self.workflow = self._build_workflow()
    
    def _build_workflow(self) -> StateGraph:
        workflow = StateGraph(AgentState)
        workflow.add_node("agent", self._agent_node)
        workflow.add_node("tools", self._tools_node)
        workflow.set_entry_point("agent")
        workflow.add_conditional_edges("agent", self._should_continue, {"continue": "tools", "end": END})
        workflow.add_conditional_edges("tools", self._after_tools, {"continue": "agent", "end": END})
        return workflow.compile()
    
    def _after_tools(self, state: AgentState) -> str:
        intermediate_steps = state.get("intermediate_steps", [])
        if not intermediate_steps:
            return "continue"
        
        # Check if the last message is a ToolMessage from find_lawyers
        last_message = intermediate_steps[-1]
        if isinstance(last_message, ToolMessage):
            if last_message.name == "_find_lawyers":
                logger.info("🛑 find_lawyers tool completed - ending with tool output")
                return "end"
        
        return "continue"
    
    def _should_continue(self, state: AgentState) -> str:
        intermediate_steps = state.get("intermediate_steps", [])
        if not intermediate_steps:
            return "continue"
        last_message = intermediate_steps[-1]
        
        if hasattr(last_message, 'tool_calls') and last_message.tool_calls:
            print(last_message.tool_calls)
            logger.info(f"🔧 Calling tools: {[tc['name'] for tc in last_message.tool_calls]}")
            return "continue"
        return "end"
    
    async def _agent_node(self, state: AgentState) -> AgentState:
        intermediate_steps = state.get("intermediate_steps", [])
        
        if not intermediate_steps:
            history = state.get("conversation_history", [])
            # Use provided system prompt if available (not None), otherwise use the default
            system_prompt_to_use = state.get("system_prompt")
            jurisdiction = state.get("jurisdiction", "unknown")
            # Deepcopy to avoid modifying the original prompt string
            system_prompt_to_use = copy.deepcopy(system_prompt_to_use)
            system_prompt_to_use = system_prompt_to_use.format(jurisdiction=jurisdiction)
            logger.info(f"📍 Formatted system prompt with jurisdiction: {jurisdiction}")
            
            intermediate_steps.append(SystemMessage(content=system_prompt_to_use))
            for msg in history:
                if isinstance(msg, dict):
                    if msg.get("role") == "user":
                        intermediate_steps.append(HumanMessage(content=msg.get("content")))
                    elif msg.get("role") == "assistant":
                        intermediate_steps.append(AIMessage(content=msg.get("content")))
            intermediate_steps.append(HumanMessage(content=state["user_query"]))
        
        response = await self.llm_with_tools.ainvoke(intermediate_steps)
        intermediate_steps.append(response)
        state["intermediate_steps"] = intermediate_steps
        return state
    
    async def _tools_node(self, state: AgentState) -> AgentState:
        logger.info("=" * 80)
        logger.info("🔧 TOOLS NODE STARTED")
        logger.info("=" * 80)
        
        intermediate_steps = state.get("intermediate_steps", [])
        last_message = intermediate_steps[-1]
        
        if not (hasattr(last_message, 'tool_calls') and last_message.tool_calls):
            logger.info("⚠️ No tool calls found in last message, skipping tools node")
            logger.info("=" * 80)
            return state
        
        logger.info(f"📋 Found {len(last_message.tool_calls)} tool call(s) to execute")
        
        for tool_call in last_message.tool_calls:
            try:
                tool_name = tool_call['name']
                logger.info(f"\n🔨 Executing tool: {tool_name}")
                logger.info(f"📝 Tool call ID: {tool_call['id']}")
                logger.info(f"📋 Tool args: {tool_call['args']}")
                
                # Use utility function to find tool with smart naming
                tool_func, tool_name_normalized = find_tool(tool_name, self.tools)
                
                if not tool_func:
                    continue
                
                logger.info(f"✅ Tool function found: {tool_func.name}")
                
                # Inject parameters from state into tool calls
                args = tool_call['args'].copy()
                
                # Use actual tool name for comparisons (includes underscore prefix)
                if tool_name_normalized in ["_find_lawyers", "_query_knowledge_graph", "_message_lawyer"]:
                    args["conversation_history"] = state.get("conversation_history", [])
                    logger.info(f"📝 Injecting conversation_history to {tool_name_normalized}: {len(args['conversation_history'])} messages")
                
                # Inject jurisdiction for query_knowledge_graph tool
                if tool_name_normalized == "_query_knowledge_graph":
                    args["jurisdiction"] = state.get("jurisdiction")
                    logger.info(f"🌍 Injecting jurisdiction: {args['jurisdiction']}")
                
                # Inject user_id for message_lawyer tool
                if tool_name_normalized == "_message_lawyer":
                    args["user_id"] = state.get("user_id")
                    logger.info(f"👤 Injecting user_id: {args['user_id']}")
                
                # Inject user_id for retrieve_lawyer_document tool
                if tool_name_normalized == "_retrieve_lawyer_document":
                    args["user_id"] = state.get("user_id")
                    logger.info(f"📄 Injecting user_id for retrieve_lawyer_document: {args['user_id']}")
                
                # Inject user_id for create_draft_document tool
                if tool_name_normalized == "_create_draft_document":
                    args["user_id"] = state.get("user_id")
                    args["instruction"] = state.get("user_query", "")
                    # Inject conversation_history for context
                    args["conversation_history"] = state.get("conversation_history", [])
                    logger.info(f"💬 Injecting conversation_history for create_draft_document: {len(args['conversation_history'])} messages")
                    # Inject doc_summaries for context
                    args["doc_summaries"] = state.get("doc_summaries", [])
                    logger.info(f"📄 Injecting doc_summaries for create_draft_document: {len(args['doc_summaries'])} documents")
                    logger.info(f"📝 Using user_query as instruction for create_draft_document")
                    logger.info(f"📝 Injecting user_id for create_draft_document: {args['user_id']}")
                
                # Use normalized name for ToolMessage
                tool_call['name'] = tool_name_normalized
                
                logger.info(f"🚀 Calling tool function.ainvoke with args: {args}")
                result = await tool_func.ainvoke(args)
                logger.info(f"✅ Tool {tool_call} returned successfully")
                logger.info(f"📊 Result type: {type(result)}")
                logger.info(f"📄 Result preview (first 500 chars): {str(result)[:500]}")
                logger.info(f"🔢 Result length: {len(str(result))} characters")
                
                intermediate_steps.append(ToolMessage(content=str(result), tool_call_id=tool_call['id'], name=tool_name_normalized))
                logger.info(f"✅ ToolMessage added to intermediate_steps")
                
            except Exception as e:
                logger.error("=" * 80)
                logger.error(f"❌ ERROR EXECUTING TOOL: {tool_name}")
                logger.error("=" * 80)
                logger.error(f"💬 Error message: {str(e)}")
                logger.error(f"🔍 Error type: {type(e).__name__}")
                logger.error(f"📋 Tool call: {tool_call}")
                logger.error(f"📄 Full traceback:")
                logger.error(traceback.format_exc())
                logger.error("=" * 80)
                
                # Add error as ToolMessage so agent knows about the failure
                error_message = f"Tool {tool_name} failed: {str(e)}"
                intermediate_steps.append(ToolMessage(
                    content=error_message,
                    tool_call_id=tool_call['id'],
                    name=tool_name_normalized
                ))
                logger.warning(f"⚠️ Error ToolMessage added to intermediate_steps, continuing workflow")
        
        state["intermediate_steps"] = intermediate_steps
        logger.info("=" * 80)
        logger.info("✅ TOOLS NODE COMPLETED")
        logger.info(f"📊 Total intermediate_steps: {len(intermediate_steps)}")
        logger.info("=" * 80)
        return state
    
    async def process_query(self, user_query: str, user_id: Optional[str] = None, jurisdiction: str = "Romania", conversation_history: Optional[List[Dict[str, str]]] = None, system_prompt: Optional[str] = None) -> Dict[str, Any]:
        initial_state = {
            "user_query": user_query,
            "user_id": user_id,
            "conversation_history": conversation_history or [],
            "intermediate_steps": [],
            "relevant_documents": [],
            "query_timestamp": datetime.now().isoformat(),
            "processing_time": None,
            "jurisdiction": jurisdiction,
            "system_prompt": system_prompt
        }
        self.performance_monitor.reset()
        
        final_state = await self.workflow.ainvoke(initial_state)
        intermediate_steps = final_state.get("intermediate_steps", [])
        final_response = intermediate_steps[-1].content
        
        logger.info("=" * 80)
        logger.info("📤 SENDING RESPONSE TO CLIENT")
        logger.info("=" * 80)
        logger.info(f"👤 User ID: {user_id}")
        logger.info(f"📅 Timestamp: {final_state.get('query_timestamp')}")
        logger.info(f"⏱️ Processing time: {sum(self.performance_monitor.get_metrics().values())}s")
        logger.info(f"📝 Response length: {len(final_response) if final_response else 0} characters")
        logger.info(f"💬 Response preview (first 200 chars): {final_response[:200] if final_response else '(empty)'}")
        logger.info(f"🔢 Full response length: {len(str(final_response)) if final_response else 0} characters")
        
        # Build the full response object
        response_obj = {
            "response": final_response or "I apologize, but I couldn't generate a response.",
            "processing_time": sum(self.performance_monitor.get_metrics().values()),
            "references": final_state.get("relevant_documents", []),
            "timestamp": final_state.get("query_timestamp")
        }
        
        logger.info(f"🔢 Full response object keys: {list(response_obj.keys())}")
        logger.info(f"📊 Response value type: {type(response_obj['response'])}")
        logger.info("=" * 80)
        logger.info(f"📤 Returning response to user")
        logger.info("=" * 80)
        
        return response_obj