"""
Identity verification plugin using InsightFace (face recognition) for MorphGuard.
"""
import os
import numpy as np
from insightface.app import FaceAnalysis
from PIL import Image

from abc import ABC, abstractmethod

class IdentityConnector(ABC):
    """Abstract interface for external identity verification connectors."""
    @abstractmethod
    def enroll(self, user_id: str, image_path: str) -> bool:
        """Enroll a user with given ID using the provided image."""
        pass

    @abstractmethod
    def verify(self, user_id: str, image_path: str) -> tuple:
        """Verify the provided image against enrolled identity; return (id, score)."""
        pass

class IdentityVerifier:
    """Enroll and verify identities against a face gallery."""
    def __init__(self, provider: str = 'insightface'):
        # Initialize InsightFace app for detection & recognition
        self.app = FaceAnalysis(allowed_modules=['detection', 'recognition'])
        self.app.prepare(ctx_id=0, det_size=(224, 224))
        # Gallery: id -> embedding
        self.gallery = {}

    def enroll(self, identity: str, image_path: str) -> bool:
        """Add a face to the gallery under given identity."""
        img = Image.open(image_path).convert('RGB')
        arr = np.array(img)
        faces = self.app.get(arr)
        if not faces:
            raise ValueError('No face detected during enrollment')
        emb = faces[0].embedding
        self.gallery[identity] = emb
        return True

    def verify(self, image_path: str) -> tuple:
        """Verify a face image against the enrolled gallery.
        Returns (best_identity, best_score) via cosine similarity.
        """
        img = Image.open(image_path).convert('RGB')
        arr = np.array(img)
        faces = self.app.get(arr)
        if not faces:
            raise ValueError('No face detected during verification')
        emb = faces[0].embedding
        best_id, best_score = None, -1.0
        for identity, g_emb in self.gallery.items():
            # Cosine similarity
            sim = float(np.dot(emb, g_emb) / (np.linalg.norm(emb) * np.linalg.norm(g_emb)))
            if sim > best_score:
                best_score, best_id = sim, identity
        return best_id, best_score