app.py

import gradio as gr
import os
import requests
import json
import shutil
from huggingface_hub import HfApi, create_repo

HF_TOKEN = os.environ.get("HF_TOKEN")

def download_file(digest, image):
    url = f"https://registry.ollama.ai/v2/library/{image}/blobs/{digest}"
    file_name = f"blobs/{digest}"

    # Create the directory if it doesn't exist
    os.makedirs(os.path.dirname(file_name), exist_ok=True)

    # Download the file
    print(f"Downloading {url} to {file_name}")
    response = requests.get(url, allow_redirects=True)
    if response.status_code == 200:
        with open(file_name, 'wb') as f:
            f.write(response.content)
    else:
        print(f"Failed to download {url}")

def fetch_manifest(image, tag):
    manifest_url = f"https://registry.ollama.ai/v2/library/{image}/manifests/{tag}"
    response = requests.get(manifest_url)
    if response.status_code == 200:
        return response.json()
    else:
        return None

def upload_to_huggingface(repo_id, folder_path, token=token):
    api = HfApi(token=token)
    repo_path = api.create_repo(repo_id=repo_id, repo_type="model", exist_ok=True)
    print(f"Repo created {repo_path}")
    try:
        api.upload_folder(
            folder_path=folder_path,
            repo_id=repo_id,
            repo_type="model",
        )
        return "Upload successful"
    except Exception as e:
        return f"Upload failed: {str(e)}"

def process_image_tag(image_tag, repo_id, oauth_token: Union[gr.OAuthToken, None]):
    # Extract image and tag from the input
    try:
        token = oauth_token.token if oauth_token else None
        
        image, tag = image_tag.split(':')
    
        # Fetch the manifest JSON
        manifest_json = fetch_manifest(image, tag)
        if not manifest_json or 'errors' in manifest_json:
            return f"Failed to fetch the manifest for {image}:{tag}"
    
        # Save the manifest JSON to the blobs folder
        manifest_file_path = "blobs/manifest.json"
        os.makedirs(os.path.dirname(manifest_file_path), exist_ok=True)
        with open(manifest_file_path, 'w') as f:
            json.dump(manifest_json, f)
    
        # Extract the digest values from the JSON
        digests = [layer['digest'] for layer in manifest_json.get('layers', [])]
    
        # Download each file
        for digest in digests:
            download_file(digest, image)
    
        # Download the config file
        config_digest = manifest_json.get('config', {}).get('digest')
        if config_digest:
            download_file(config_digest, image)
    
        # Upload to Hugging Face Hub
        upload_result = upload_to_huggingface(repo_id, 'blobs/*', token=token)
    
        # Delete the blobs folder
        shutil.rmtree('blobs')
        return f"Successfully fetched and downloaded files for {image}:{tag}\n{upload_result}\nBlobs folder deleted"
    except Exception as e:
        shutil.rmtree('blobs', ignore_errors=True)
        return f"Error found: {str(e)}"

css = """
.main_ui_logged_out{opacity: 0.3; pointer-events: none}
"""

# Create the Gradio interface using gr.Blocks
with gr.Blocks(css=css) as demo:
    gr.Markdown("# Ollama <> HF Hub 🤝")
    gr.Markdown("Enter the image and tag to download the corresponding files from the Ollama registry and upload them to the Hugging Face Hub.")
    
    gr.LoginButton()
    
    image_tag_input = gr.Textbox(placeholder="Enter Ollama ID", label="Image and Tag")
    repo_id_input = gr.Textbox(placeholder="Enter Hugging Face repo ID", label="Hugging Face Repo ID")
    
    result_output = gr.Textbox(label="Result")
    
    process_button = gr.Button("Process")
    process_button.click(fn=process_image_tag, inputs=[image_tag_input, repo_id_input], outputs=result_output)

# Launch the Gradio app
demo.launch()