app.py

import os, io, base64
from typing import List, Dict, Any
import gradio as gr
from PIL import Image

from car_core.specs_model_space import build_label_space
from car_core.models_zeroshot import ModelIDZeroShot
from car_core.color_detect import dominant_color
from car_core.issues import detect_issues
from car_core.pricing import price_issues, load_regions
from car_core.exporter import export_pdf, export_json

LABEL_SPACE = build_label_space()
MODEL = ModelIDZeroShot(LABEL_SPACE)
REGIONS = load_regions()

def _to_pil(obj):
    if isinstance(obj, dict) and "image" in obj:
        # Gradio File with base64 'image' key
        return Image.open(io.BytesIO(base64.b64decode(obj["image"].split(",")[-1])))
    if isinstance(obj, str):
        return Image.open(obj)
    if hasattr(obj, "read"):
        return Image.open(obj)
    return obj

def analyze(images: list, region: str):
    if not images:
        raise gr.Error("Upload at least one car image.")
    imgs = []
    for it in images:
        try:
            imgs.append(_to_pil(it))
        except Exception:
            pass
    if not imgs:
        raise gr.Error("Failed to decode images. Use JPG/PNG.")

    # Model decision: pick the most frequent top label across images
    votes = {}
    for im in imgs:
        lbl = MODEL.top_label(im)
        votes[lbl] = votes.get(lbl, 0) + 1
    model_final = sorted(votes.items(), key=lambda kv: kv[1], reverse=True)[0][0]

    # Color decision: take the most frequent named color across images
    color_votes = {}
    for im in imgs:
        c = dominant_color(im)["name"]
        color_votes[c] = color_votes.get(c, 0) + 1
    color_name = sorted(color_votes.items(), key=lambda kv: kv[1], reverse=True)[0][0]
    color_any = None
    # Recompute once to get rgb/hex for the chosen name
    for im in imgs:
        d = dominant_color(im)
        if d["name"] == color_name:
            color_any = d; break
    color_final = color_any or {"name": color_name, "rgb": (0,0,0), "hex": "#000000"}

    # Issues (deterministic)
    issues = detect_issues(imgs)

    # Pricing
    pricing = price_issues(issues, region_code=region)

    payload = {
        "vehicle": {"model": model_final, "color": color_final},
        "region": region,
        "issues": issues,
        "pricing": pricing
    }

    os.makedirs("exports", exist_ok=True)
    pdf_path = "exports/report.pdf"
    json_path = "exports/report.json"
    export_pdf(payload, pdf_path)
    export_json(payload, json_path)

    def to_dl(path):
        with open(path, "rb") as f:
            return (os.path.basename(path), f.read())

    # Deterministic, actionable output only
    result = {
        "vehicle": payload["vehicle"],
        "region": pricing["region"],
        "currency": pricing["currency"],
        "issues_with_solutions": [
            {
                "issue": it["issue"],
                "solution": it["solution"],
                "labor_hours": it["labor_hours"],
                "labor_cost": it["labor_cost"],
                "parts_cost": it["parts_cost"],
                "line_total": it["line_total"]
            } for it in pricing["items"]
        ],
        "totals": {
            "subtotal": pricing["subtotal"],
            "tax": pricing["tax"],
            "grand_total": pricing["grand_total"]
        }
    }
    return result, to_dl(pdf_path), to_dl(json_path)

with gr.Blocks(fill_height=True) as demo:
    gr.Markdown("## 🛠️ Car Analysis Advisor — multi‑image model/color/issue detection with exact pricing")
    with gr.Row():
        with gr.Column(scale=1):
            imgs = gr.File(label="Upload car image(s)", file_count="multiple", file_types=["image"])
            region = gr.Dropdown(choices=list(REGIONS["regions"].keys()), value=REGIONS.get("default_region","IN-HYD"), label="Region (pricing)")
            run = gr.Button("Analyze", variant="primary")
        with gr.Column(scale=1):
            out = gr.JSON(label="Actionable results")
            pdf = gr.File(label="Download PDF")
            jj = gr.File(label="Download JSON")
    run.click(analyze, inputs=[imgs, region], outputs=[out, pdf, jj])

if __name__ == "__main__":
    demo.launch()