#!/usr/bin/env python3
# Object Detection (concise, correct Content-Type). Requires: pillow, requests, config.py (HF_API_KEY).
import os, io, time, random, requests, mimetypes
from datetime import datetime
from PIL import Image, ImageDraw, ImageFont
from config import HF_API_KEY

MODEL = "facebook/detr-resnet-50"
API = f"https://[Log in to view URL]"
ALLOWED, MAX_MB = {".jpg",".jpeg",".png",".bmp",".gif",".webp",".tiff"}, 8
EMOJI = {"person":"🧍","car":"🚗","truck":"🚚","bus":"🚌","bicycle":"🚲","motorcycle":"🏍️","dog":"🐶","cat":"🐱",
         "bird":"🐦","horse":"🐴","sheep":"🐑","cow":"🐮","bear":"🐻","giraffe":"🦒","zebra":"🦓","banana":"🍌",
         "apple":"🍎","orange":"🍊","pizza":"🍕","broccoli":"🥦","book":"📘","laptop":"💻","tv":"📺","bottle":"🧴","cup":"🥤"}

def font(sz=18):
    for f in ("DejaVuSans.ttf","arial.ttf"):
        try: return ImageFont.truetype(f, sz)
        except: pass
    return ImageFont.load_default()

def ask_image():
    print("\n🎯 Pick an image (JPG/PNG/WebP/BMP/TIFF <= 8MB) from this folder.")
    while True:
        p = input("Image path: ").strip().strip('"').strip("'")
        if not p or not os.path.isfile(p): print("⚠️ Not found."); continue
        if os.path.splitext(p)[1].lower() not in ALLOWED: print("⚠️ Unsupported type."); continue
        if os.path.getsize(p)/(1024*1024) > MAX_MB: print("⚠️ Too big (>8MB)."); continue
        try: Image.open(p).verify()
        except: print("⚠️ Corrupted image."); continue
        return p

def infer(path, img_bytes, tries=8):
    mime, _ = mimetypes.guess_type(path)
    for _ in range(tries):
        if mime and mime.startswith("image/"):
            r = requests.post(API,
                headers={"Authorization": f"Bearer {HF_API_KEY}",
                         "Content-Type": mime},
                data=img_bytes, timeout=60)
        else:
            r = requests.post(API,
                headers={"Authorization": f"Bearer {HF_API_KEY}"},
                files={"inputs": (os.path.basename(path), img_bytes, "application/octet-stream")},
                timeout=60)
        if r.status_code == 200:
            d = r.json()
            if isinstance(d, dict) and "error" in d: raise RuntimeError(d["error"])
            if not isinstance(d, list): raise RuntimeError("Bad API response.")
            return d
        if r.status_code == 503: time.sleep(2); continue
        raise RuntimeError(f"API {r.status_code}: {r.text[:300]}")
    raise RuntimeError("Model warm-up timeout.")

def draw(img, dets, thr=0.5):
    d = ImageDraw.Draw(img); f = font(18); counts = {}
    for det in dets[:50]:
        s = float(det.get("score",0)); 
        if s < thr: continue
        lab = det.get("label","object"); b = det.get("box",{})
        x1,y1,x2,y2 = (int(b.get(k,0)) for k in ("xmin","ymin","xmax","ymax"))
        if not (x2>0 and y2>0):
            x,y,w,h = int(b.get("x",0)), int(b.get("y",0)), int(b.get("w",0)), int(b.get("h",0))
            x1,y1,x2,y2 = x,y,x+w,y+h
        color = tuple(random.randint(80,255) for _ in range(3))
        d.rectangle([(x1,y1),(x2,y2)], outline=color, width=4)
        txt = f"{EMOJI.get(lab.lower(),'✨')} {lab} {s*100:.0f}%"
        tw,th = d.textlength(txt,font=f), f.size+6
        d.rectangle([(x1,max(0,y1-th)),(x1+tw+8,y1)], fill=color)
        d.text((x1+4,y1-th+3), txt, font=f, fill=(0,0,0))
        counts[lab] = counts.get(lab,0)+1
    return counts

def main():
    path = ask_image()
    with open(path,"rb") as fh: by = fh.read()
    try: dets = infer(path, by)
    except Exception as e: return print("❌", e)
    img = Image.open(io.BytesIO(by)).convert("RGB")
    counts = draw(img, dets, thr=0.5)
    out = f"annotated_{datetime.now().strftime('%Y%m%d_%H%M%S')}.png"
    img.save(out); print(f"✅ Saved: {out}")
    if counts:
        print("🎉 I found:"); 
        for k,v in sorted(counts.items(), key=lambda kv:(-kv[1],kv[0])):
            print(f" • {EMOJI.get(k.lower(),'✨')} {k}: {v}")
    else:
        print("🤔 No confident detections—try a clearer or busier scene!")
    
    print("\n⚠️ Disclaimer: This is an AI model demo. "
          "Detections may not always be accurate or complete. "
          "Use it for fun and learning, not for safety-critical decisions.")

if __name__=="__main__": main()