app.py

import gradio as gr
from huggingface_hub import HfApi
from git import Repo
from slugify import slugify
import uuid
import os
import re

# Function to extract metadata from README.md file
def extract_from_readme(readme_path):
    title = None
    description = None
    port = None
    if os.path.exists(readme_path):
        with open(readme_path, encoding="utf-8") as f:
            content = f.read()
            match_title = re.search(r'title:\s*(.*)', content)
            match_desc = re.search(r'short_description:\s*(.*)', content)
            match_port = re.search(r'app_port:\s*(\d+)', content)
            if match_title:
                title = match_title.group(1).strip()
            if match_desc:
                description = match_desc.group(1).strip()
            if match_port:
                port = match_port.group(1).strip()
    return title, description, port

# Main function to clone a Git repo and create a Hugging Face repo
def clone_and_create(profile: gr.OAuthProfile, token: gr.OAuthToken,
                     repo_url, repo_type, private, space_sdk, sdk_version,
                     space_id, space_title, space_description):

    # README.md から情報を抽出する関数
    def extract_from_readme(readme_path):
        import re
        title = None
        description = None
        port = None
        if os.path.exists(readme_path):
            with open(readme_path, encoding="utf-8") as f:
                content = f.read()
                match_title = re.search(r'title:\s*(.*)', content)
                match_desc = re.search(r'short_description:\s*(.*)', content)
                match_port = re.search(r'app_port:\s*(\d+)', content)
                if match_title:
                    title = match_title.group(1).strip()
                if match_desc:
                    description = match_desc.group(1).strip()
                if match_port:
                    port = match_port.group(1).strip()
        return title, description, port

    folder = str(uuid.uuid4())  # ユニークなフォルダ名作成
    Repo.clone_from(repo_url, folder)  # Gitリポジトリをクローン

    hf_api = HfApi(token=token.token)

    # README.md から自動情報取得
    title_from_readme, desc_from_readme, port_from_readme = extract_from_readme(os.path.join(folder, "README.md"))

    # 最終的なID, タイトル, 説明を決定
    final_id = slugify(space_id) if space_id else slugify(repo_url.split("/")[-1])
    final_title = space_title or title_from_readme or final_id
    final_description = space_description or desc_from_readme
    repo_visibility = "private" if private else "public"

    # リポジトリ作成時の引数設定
    if repo_type == "space":
        create_kwargs = dict(
            repo_id=f"{profile.username}/{final_id}",
            repo_type="space",
            space_sdk=space_sdk,
            private=private,
        )
        # gradio, streamlit の場合のみバージョンを渡す
        if space_sdk in ["gradio", "streamlit"] and sdk_version:
            create_kwargs["space_sdk_version"] = sdk_version

        hf_api.create_repo(**create_kwargs)
    else:
        hf_api.create_repo(
            repo_id=f"{profile.username}/{final_id}",
            repo_type=repo_type,
            private=private
        )

    # README.md に不足情報を追記
    readme_path = os.path.join(folder, "README.md")
    with open(readme_path, "a", encoding="utf-8") as f:
        if space_title and not title_from_readme:
            f.write(f"\ntitle: {space_title}")
        if space_description and not desc_from_readme:
            f.write(f"\nshort_description: {space_description}")
        if port_from_readme:
            f.write(f"\napp_port: {port_from_readme}")

    # フォルダ内容をHugging Faceにアップロード
    hf_api.upload_folder(
        folder_path=folder,
        repo_id=f"{profile.username}/{final_id}",
        repo_type=repo_type
    )

    # 作成したURLを返す
    base_url = "spaces" if repo_type == "space" else (repo_type + "s" if repo_type != "model" else "models")
    return f"https://huggingface.co/{base_url}/{profile.username}/{final_id}"


# Define Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# 🚀 Git Repository → Hugging Face Clone Tool")
    gr.LoginButton()

    with gr.Row():
        with gr.Column():
            repo_url = gr.Textbox(label="Git Repository URL (e.g., GitHub)")
            repo_type = gr.Dropdown(choices=["space", "dataset", "model"], value="space", label="Hugging Face repo type")
            private = gr.Checkbox(label="Make Private", value=False)
            space_sdk = gr.Dropdown(choices=["gradio", "streamlit", "docker", "static"], label="Space SDK", visible=True)
            sdk_version = gr.Textbox(label="SDK Version (optional)", visible=True)
            space_id = gr.Textbox(label="Hugging Face ID (auto if blank)")
            space_title = gr.Textbox(label="Space Title (from README or ID if blank)")
            space_description = gr.Textbox(label="Space Description (from README if blank)")

        with gr.Column():
            output = gr.Textbox(label="Created Hugging Face URL")

    # Show/hide some inputs depending on repo_type
    def toggle_space_fields(repo_type_val):
        visible = repo_type_val == "space"
        return {
            space_sdk: gr.update(visible=visible),
            sdk_version: gr.update(visible=visible),
            space_id: gr.update(visible=True),
            space_title: gr.update(visible=True),
            space_description: gr.update(visible=True),
        }

    repo_type.change(fn=toggle_space_fields, inputs=repo_type, outputs=[space_sdk, sdk_version, space_id, space_title, space_description])

    btn = gr.Button("Clone and Create!")
    btn.click(
        fn=clone_and_create,
        inputs=[repo_url, repo_type, private, space_sdk, sdk_version, space_id, space_title, space_description],
        outputs=output
    )

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