app.py

import tempfile
import warnings
import pickle
from pathlib import Path

import gradio as gr
import joblib
from skops import io as sio

title = "skops converter"

desc = """
# Pickle to skops converter

This space converts your pickle files to skops format. You can read more on the
skops format [here]( https://skops.readthedocs.io/en/stable/persistence.html).

You can use `skops.io.dump(joblib.load(in_file), out_file)` to do the
conversion yourself, where `in_file` is your source pickle file and `out_file`
is where you want to save the skops file. But only do that **if you trust the
source of the pickle file**.

You can then use `skops.io.load(skops_file, trusted=unknown_types)` to load the
file, where `skops_file` is the converted skops format file, and the
`unknown_types` is what you see in the "Unknown Types" box bellow. You can also
locally reproduce this list using
`skops.io.get_untrusted_types(file=skops_file)`. You should only load a `skops`
file that you trust all the types included in the `unknown_types` list.

## Requirements

This space assumes you have used the latest `joblib` and `scikit-learn`
versions installed on your environment to create the pickle file.

## Reporting issues

If you encounter an issue, please open an issue on the project's repository
on the [issue tracker](
https://github.com/skops-dev/skops/issues/new?title=CONVERSION+error+from+hf.space&body=Paste+the+error+message+and+a+link+to+your+pickle+file+here+please)

"""


def convert(file):
    msg = ""
    try:
        with warnings.catch_warnings(record=True) as record:
            in_file = Path(file.name)
            try:
                obj = joblib.load(in_file)
            except:
                with open(in_file, "rb") as f:
                    obj = pickle.load(f)

            if "." in in_file.name:
                out_file = ".".join(in_file.name.split(".")[:-1])
            else:
                out_file = in_file.name

            out_file += ".skops"
            path = tempfile.mkdtemp(prefix="gradio-convert-")
            out_file = Path(path) / out_file
            sio.dump(obj, out_file)
            unknown_types = sio.get_untrusted_types(file=out_file)
        if len(record):
            msg = "\n".join([repr(w.message) for w in record])
    except Exception as e:
        return None, None, repr(e)
    return out_file, unknown_types, msg


with gr.Blocks(title=title) as iface:
    gr.Markdown(desc)
    upload_button = gr.UploadButton(
        "Click to Upload a File",
        file_types=None,
        file_count="single",
    )
    file_output = gr.File(label="Converted File")
    upload_button.upload(
        convert,
        upload_button,
        [
            file_output,
            gr.Text(label="Unknown Types"),
            gr.Text(label="Errors and Warnings"),
        ],
        api_name="upload-file"
    )


iface.launch(debug=True)