"""
Protocol 29: Antigravity Context Oracle Tools
These tools allow Agents to interact with the Prime-Neuron Context Service.
"""
import requests
import json
from logos.config import SERVER_URL

def query_prime_context(query_text: str, prime_band: str = None) -> str:
    """
    Semantic search over the Prime Manifold.
    Args:
        query_text: Natural language query
        prime_band: "axioms" (0-1000), "mid" (1000-5000), "hitech" (5000+)
    """
    url = f"{SERVER_URL}/v1/context/query"
    
    filters = {}
    if prime_band:
        if prime_band == "axioms": filters["prime_range"] = [0, 1000]
        elif prime_band == "mid": filters["prime_range"] = [1000, 5000]
        elif prime_band == "hitech": filters["prime_range"] = [5000, 99999]
        
    try:
        resp = requests.post(url, json={"query_text": query_text, "filters": filters})
        if resp.status_code == 200:
            data = resp.json()
            # Synthesize for Agent
            summary = [f"Found {data['count']} neurons:"]
            for n in data['results']:
                summary.append(f"- [Prime {n.get('prime_index')}] {n.get('type')}: {str(n.get('payload'))[:50]}...")
            return "\n".join(summary)
        else:
            return f"Error: {resp.text}"
    except Exception as e:
        return f"Connection Failed: {e}"

def upsert_prime_neuron(content: str, type: str = "text", prime_index: int = None) -> str:
    """
    Writes a new concept to the Manifold.
    """
    url = f"{SERVER_URL}/v1/context/neurons"
    
    neuron = {
        "type": type,
        "payload": content
    }
    if prime_index:
        neuron["prime_index"] = prime_index
        
    try:
        resp = requests.post(url, json={"neurons": [neuron]})
        if resp.status_code == 200:
            data = resp.json()
            params = data['neurons'][0]
            return f"Upserted Neuron: ID={params.get('id')} Prime={params.get('prime_index')}"
        else:
            return f"Error: {resp.text}"
    except Exception as e:
        return f"Connection Failed: {e}"

def parse_diagram_to_context(image_path: str, domain_context: str = "General") -> str:
    """
    Ingests a diagram image, segments it, and stores nodes as neurons.
    Currently a stub for Protocol 29 Step 3.
    """
    if not image_path: return "Error: No image path provided."
    
    # In a real implementation, this would:
    # 1. Call Local Vision Model (Ollama/Llava) to describe image
    # 2. Parse graph structure
    # 3. Upsert nodes
    
    # Simulating a mock upsert for the uploaded diagram
    mock_neuron_text = f"Diagram Node from {image_path}: {domain_context}"
    return upsert_prime_neuron(mock_neuron_text, "diagram_node")

if __name__ == "__main__":
    # Test
    print("Upserting...")
    print(upsert_prime_neuron("The LOGOS System requires manifold constraints.", "axiom", 2))
    print("Querying...")
    print(query_prime_context("manifold"))