from __future__ import annotations
from typing import Optional, Dict, Tuple
from PIL import Image
import numpy as np
import cv2
import os

def pil_from_path(path: str) -> Optional[Image.Image]:
    try:
        return Image.open(path).convert("RGB")
    except Exception:
        return None

def first_frame(path: str, max_side: int = 960) -> Tuple[Optional[Image.Image], Dict[str, float]]:
    info: Dict[str, float] = {}
    try:
        cap = cv2.VideoCapture(path)
        if not cap.isOpened():
            return None, {"error": "Cannot open video"}
        fps = cap.get(cv2.CAP_PROP_FPS) or 0.0
        w = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH) or 0)
        h = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT) or 0)
        frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT) or 0)
        ok, frame = cap.read()
        cap.release()
        if not ok or frame is None:
            return None, {"error": "Failed to read first frame"}
        frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
        scale = min(1.0, max_side / max(h, w)) if max(h, w) > 0 else 1.0
        if scale < 1.0:
            frame = cv2.resize(frame, (int(w*scale), int(h*scale)), interpolation=cv2.INTER_AREA)
        dur = (frames / fps) if fps > 0 else 0.0
        return Image.fromarray(frame), {"width": w, "height": h, "fps": round(fps,3), "frames": frames, "duration_s": round(dur,2)}
    except Exception as e:
        return None, {"error": str(e)}

def mask_debug_on_image(img: Image.Image) -> Image.Image:
    """Classical single-frame mask heuristic for quick sanity checks (no models)."""
    ar = np.array(img)
    if ar.ndim == 3 and ar.shape[2] == 3:
        gray = cv2.cvtColor(ar, cv2.COLOR_RGB2GRAY)
    else:
        gray = ar if ar.ndim == 2 else cv2.cvtColor(ar, cv2.COLOR_RGBA2GRAY)

    edges = cv2.Canny(gray, 80, 160)
    edges = cv2.dilate(edges, cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3)), 1)

    border = np.concatenate([gray[0, :], gray[-1, :], gray[:, 0], gray[:, -1]])
    bg_median = np.median(border)
    diff = np.abs(gray.astype(np.float32) - bg_median)
    thresh = (diff > 28).astype(np.uint8) * 255

    mask = cv2.bitwise_or(thresh, edges)
    k7 = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (7, 7))
    mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, k7)
    mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, k7)
    mask_c = cv2.cvtColor(mask, cv2.COLOR_GRAY2RGB)
    overlay = (0.6 * ar + 0.4 * np.dstack([mask, np.zeros_like(mask), np.zeros_like(mask)])).astype(np.uint8)
    return Image.fromarray(overlay)