from mcp.server.fastmcp import FastMCP
import logging
import os

# Initialize the server
mcp = FastMCP("Math-Education-Server")

current_dir = os.path.dirname(os.path.abspath(__file__))

# Path to your collected resource
MARKDOWN_FILE = r"resource\jemh114-min (1).md" 

def get_local_resource():
    """Helper to read the markdown file safely."""
    if os.path.exists(MARKDOWN_FILE):
        with open(MARKDOWN_FILE, "r", encoding="utf-8") as f:
            return f.read()
    return "Resource file not found."

# --- Task 1: Generate a Summary ---
@mcp.tool()
async def generate_chapter_summary(chapter_name: str) -> str:
    """
    Provides the source material for a chapter. 
    The AI should use this to create a summary.
    """
    raw_data = get_local_resource()
    
    # We use XML-style tags. Senior engineers do this because 
    # LLMs (especially Claude) are trained to handle tagged data perfectly.
    return f"""
    <source_material>
    {raw_data[:4000]} 
    </source_material>

    SYSTEM_INSTRUCTION: You are a Mathematics Expert. 
    1. Read the text inside <source_material>.
    2. Create a summary with 3 bullet points.
    3. List all formulas found in LaTeX format.
    4. Do not repeat the raw text; only provide the summary.
    """

# --- Task 2: Quiz Generator ---
@mcp.tool()
async def generate_quiz(chapter_name: str, difficulty: str = "medium") -> str:
    """
    Extracts formulas from the markdown and asks the LLM to build a quiz.
    """
    raw_data = get_local_resource()
    
    # Using the LLM's ability to pick out formulas from the raw text provided
    return (
        f"Here is the raw material for {chapter_name}:\n"
        f"{raw_data}\n"
        f"Difficulty: {difficulty}\n"
        "Task: Create a 3-question quiz using the formulas found in the text."
    )

if __name__ == "__main__":
    # FastMCP handles the --transport sse argument automatically 
    # if you use mcp.run()
    mcp.run()