"""Simple agent implementation using LangGraph."""

import sys
from pathlib import Path
from typing import Annotated, TypedDict

from langchain_core.messages import BaseMessage, HumanMessage, SystemMessage
from langchain_openai import ChatOpenAI
from langgraph.graph import StateGraph, START, END
from langgraph.graph.message import add_messages

# Add parent directory to path for imports
sys.path.insert(0, str(Path(__file__).parent.parent))

from config.settings import get_settings
from utils.markdown_loader import load_agent_definition, get_system_prompt, load_process_specifications
from tools.summarize import summarize_text


class AgentState(TypedDict):
    """State schema for the agent graph."""

    messages: Annotated[list[BaseMessage], add_messages]
    agent_name: str


def create_agent_graph(agent_definition_path: Path | None = None):
    """
    Create a LangGraph agent from a markdown definition.
    
    Args:
        agent_definition_path: Path to the agent definition markdown file.
                              If None, uses default path.
    
    Returns:
        Compiled LangGraph graph
    """
    settings = get_settings()
    
    # Load agent definition
    if agent_definition_path is None:
        agent_definition_path = settings.docs_dir / "agent_definition.md"
    
    agent_def = load_agent_definition(agent_definition_path)
    
    # Load process specifications
    process_description, process_constraints = load_process_specifications(settings.docs_dir)
    
    # Generate system prompt with process context
    system_prompt = get_system_prompt(
        agent_def,
        process_description=process_description,
        process_constraints=process_constraints
    )
    
    # Initialize LLM with tools
    llm = ChatOpenAI(
        model=settings.openai_model,
        temperature=settings.temperature,
        max_tokens=settings.max_tokens,
        api_key=settings.openai_api_key,
    )
    
    # Bind tools to the LLM
    tools = [summarize_text]
    llm_with_tools = llm.bind_tools(tools)
    
    # Define agent node
    def agent_node(state: AgentState) -> AgentState:
        """Process messages through the agent."""
        messages = state["messages"]
        
        # Prepend system message if not already present
        if not messages or not isinstance(messages[0], SystemMessage):
            messages = [SystemMessage(content=system_prompt)] + messages
        
        # Get response from LLM with tools
        response = llm_with_tools.invoke(messages)
        
        return {
            "messages": [response],
            "agent_name": agent_def.get("name", "Assistant Agent"),
        }
    
    # Define tool execution node
    def tool_node(state: AgentState) -> AgentState:
        """Execute tools if requested by the agent."""
        from langgraph.prebuilt import ToolNode
        tool_executor = ToolNode(tools)
        return tool_executor.invoke(state)
    
    # Routing function to decide if we should use tools
    def should_continue(state: AgentState) -> str:
        """Determine if we should continue to tools or end."""
        messages = state["messages"]
        last_message = messages[-1]
        
        # If there are tool calls, route to tools
        if hasattr(last_message, "tool_calls") and last_message.tool_calls:
            return "tools"
        # Otherwise, end
        return "end"
    
    # Build the graph
    workflow = StateGraph(AgentState)
    
    # Add nodes
    workflow.add_node("agent", agent_node)
    workflow.add_node("tools", tool_node)
    
    # Add edges
    workflow.add_edge(START, "agent")
    workflow.add_conditional_edges(
        "agent",
        should_continue,
        {
            "tools": "tools",
            "end": END,
        }
    )
    workflow.add_edge("tools", "agent")
    
    # Compile the graph
    graph = workflow.compile()
    
    return graph


def run_agent(user_input: str, agent_definition_path: Path | None = None) -> str:
    """
    Run the agent with a single user input.
    
    Args:
        user_input: User's message
        agent_definition_path: Optional path to agent definition
        
    Returns:
        Agent's response as a string
    """
    graph = create_agent_graph(agent_definition_path)
    
    # Create initial state
    initial_state = {
        "messages": [HumanMessage(content=user_input)],
        "agent_name": "Assistant Agent",
    }
    
    # Run the graph
    result = graph.invoke(initial_state)
    
    # Extract the last message (agent's response)
    last_message = result["messages"][-1]
    return last_message.content

# Made with Bob