app_gradio/app.py

"""Provide an image of a chessboard and get the FEN (https://en.wikipedia.org/wiki/Forsyth–Edwards_Notation ) representation of the board."""
import argparse
import json
import logging
import os
from pathlib import Path
from typing import Callable

import gradio as gr
from PIL import ImageStat
from PIL.Image import Image
import requests

from image_to_fen.fen import ImageToFen, fen_and_image
import image_to_fen.util as util

os.environ["CUDA_VISIBLE_DEVICES"] = ""  # do not use GPU

logging.basicConfig(level=logging.INFO)

APP_DIR = Path(__file__).resolve().parent
FAVICON = APP_DIR / "265f.png"
README = APP_DIR / "README.md"

DEFAULT_PORT = 11700


def main(args):
    predictor = PredictorBackend(url=args.model_url)
    frontend = make_frontend(
        predictor.run
    )
    frontend.launch()


def make_frontend(
    fn: Callable[[Image], str],
    app_name: str = "image-to-fen"
):
    """Creates a gradio.Interface frontend for an image to text function."""
    examples_dir = Path("image_to_fen") / "tests" / "support" / "boards"
    example_fnames = [elem for elem in os.listdir(examples_dir) if elem.endswith(".png")]
    example_paths = [examples_dir / fname for fname in example_fnames]
    examples = [[str(path)] for path in example_paths]

    allow_flagging = "never"

    readme = _load_readme(with_logging=allow_flagging == "manual")

    # build a basic browser interface to a Python function
    frontend = gr.Interface(
        fn=fn,
        outputs=[gr.components.Textbox(), gr.components.Image()],
        inputs=gr.components.Image(type="pil", label="Chess Board"),
        title="♟️ Image to Fen",
        thumbnail="FAVICON",
        description=__doc__,
        article=readme,
        examples=examples,
        cache_examples=False,
        allow_flagging=allow_flagging
    )

    return frontend


class PredictorBackend:
    """Interface to a backend that serves predictions.

    To communicate with a backend accessible via a URL, provide the url kwarg.

    Otherwise, runs a predictor locally.
    """

    def __init__(self, url=None):
        if url is not None:
            self.url = url
            self._predict = self._predict_from_endpoint
        else:
            model = ImageToFen()
            # self._predict = model.predict
            self._predict = fen_and_image

    def run(self, image):
        pred, metrics = self._predict_with_metrics(image)
        self._log_inference(pred, metrics)
        return pred

    def _predict_with_metrics(self, image):
        pred = self._predict(image)

        stats = ImageStat.Stat(image)
        metrics = {
            "image_mean_intensity": stats.mean,
            "image_median": stats.median,
            "image_extrema": stats.extrema,
            "image_area": image.size[0] * image.size[1],
            "pred_length": len(pred),
        }
        return pred, metrics

    def _predict_from_endpoint(self, image):
        """Send an image to an endpoint that accepts JSON and return the predicted text.

        The endpoint should expect a base64 representation of the image, encoded as a string,
        under the key "image". It should return the predicted text under the key "pred".

        Parameters
        ----------
        image
            A PIL image of a chess board.

        Returns
        -------
        pred
            A string containing the predictor's guess of the FEN representation of the chess board.
        """
        encoded_image = util.encode_b64_image(image)

        headers = {"Content-type": "application/json"}
        payload = json.dumps({"image": "data:image/png;base64," + encoded_image})

        response = requests.post(self.url, data=payload, headers=headers)
        pred = response.json()["pred"]

        return pred

    def _log_inference(self, pred, metrics):
        for key, value in metrics.items():
            logging.info(f"METRIC {key} {value}")
        logging.info(f"PRED >begin\n{pred}\nPRED >end")


def _load_readme(with_logging=False):
    with open(README) as f:
        lines = f.readlines()
        # if not with_logging:
        #     lines = lines[: lines.index("<!-- logging content below -->\n")]

        readme = "".join(lines)
    return readme


def _make_parser():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument(
        "--model_url",
        default=None,
        type=str,
        help="Identifies a URL to which to send image data. Data is base64-encoded, converted to a utf-8 string, and then set via a POST request as JSON with the key 'image'. Default is None, which instead sends the data to a model running locally.",
    )
    parser.add_argument(
        "--port",
        default=DEFAULT_PORT,
        type=int,
        help=f"Port on which to expose this server. Default is {DEFAULT_PORT}.",
    )

    return parser


if __name__ == "__main__":
    parser = _make_parser()
    args = parser.parse_args()
    main(args)