app.py

import gradio as gr
import os
import tempfile
import shutil
import re
import json
import datetime
from pathlib import Path
from huggingface_hub import HfApi, hf_hub_download, snapshot_download
from safetensors.torch import load_file
import torch
import subprocess

# --- Conversion Function: Safetensors (UNet) → GGUF ---
def convert_unet_to_gguf(safetensors_path, output_dir, progress=gr.Progress()):
    """
    Converts a UNet safetensors file to GGUF using gguf-connector's CLI (t2 or t).
    Assumes the file is named 'unet.safetensors'.
    """
    progress(0.1, desc="Starting UNet to GGUF conversion...")

    try:
        # Ensure gguf-connector is available
        import gguf_connector  # noqa

        # Copy input to working dir because ggc expects files in current dir
        work_dir = tempfile.mkdtemp()
        input_path = os.path.join(work_dir, "unet.safetensors")
        shutil.copy(safetensors_path, input_path)

        # GGUF output will be named automatically like unet.safetensors -> unet.gguf
        gguf_output_path = os.path.join(work_dir, "unet.gguf")

        progress(0.3, desc="Running gguf-connector (t2: safetensors → GGUF)...")

        # Use 'ggc t2' for conversion (beta: unlimited)
        # This is interactive, so we must simulate input via echo or expect
        # But since ggc t2 may be interactive, we try non-interactive fallback:
        # Unfortunately, ggc does not support non-interactive mode robustly.
        # So we simulate by running in dir and hoping it picks the only file.

        # Change working dir so ggc sees the file
        original_cwd = os.getcwd()
        os.chdir(work_dir)

        try:
            # Launch ggc t2 and auto-select first file via input redirection
            # This is fragile but best-effort
            result = subprocess.run(
                ["ggc", "t2"],
                input="1\n",  # select first model
                text=True,
                capture_output=True,
                timeout=300
            )
            if result.returncode != 0:
                raise RuntimeError(f"ggc t2 failed: {result.stderr}")
        finally:
            os.chdir(original_cwd)

        if not os.path.exists(gguf_output_path):
            # Try alternative naming
            candidates = [f for f in os.listdir(work_dir) if f.endswith(".gguf")]
            if not candidates:
                raise FileNotFoundError("No GGUF file generated by ggc t2")
            gguf_output_path = os.path.join(work_dir, candidates[0])

        # Move to output dir
        final_gguf_path = os.path.join(output_dir, "unet.gguf")
        shutil.move(gguf_output_path, final_gguf_path)

        # Also save minimal config
        config_path = os.path.join(output_dir, "config.json")
        with open(config_path, "w") as f:
            json.dump({
                "model_type": "unet",
                "format": "gguf",
                "source": "converted from safetensors"
            }, f)

        progress(1.0, desc="Conversion to GGUF complete!")
        return True, "UNet converted to GGUF successfully."

    except Exception as e:
        return False, str(e)
    finally:
        if 'work_dir' in locals():
            shutil.rmtree(work_dir, ignore_errors=True)

# --- Main Processing Function ---
def process_and_upload_unet_to_gguf(repo_url, hf_token, new_repo_id, private_repo, progress=gr.Progress()):
    if not all([repo_url, hf_token, new_repo_id]):
        return None, "❌ Error: Please fill in all fields.", ""

    if not re.match(r"^[a-zA-Z0-9._-]+/[a-zA-Z0-9._-]+$", new_repo_id):
        return None, "❌ Error: Invalid repository ID format. Use 'username/model-name'.", ""

    temp_dir = tempfile.mkdtemp()
    output_dir = tempfile.mkdtemp()

    try:
        # Authenticate
        progress(0.05, desc="Logging into Hugging Face...")
        api = HfApi(token=hf_token)
        user_info = api.whoami()
        user_name = user_info['name']
        progress(0.1, desc=f"Logged in as {user_name}.")

        # Parse source repo
        clean_url = repo_url.strip().rstrip("/")
        if "huggingface.co" not in clean_url:
            return None, "❌ Source must be a Hugging Face model repo.", ""
        src_repo_id = clean_url.replace("https://huggingface.co/", "")

        # Download only unet.safetensors
        progress(0.15, desc="Downloading unet.safetensors...")
        safetensors_path = hf_hub_download(
            repo_id=src_repo_id,
            filename="unet.safetensors",
            cache_dir=temp_dir,
            token=hf_token
        )
        progress(0.3, desc="Download complete.")

        # Convert
        success, msg = convert_unet_to_gguf(safetensors_path, output_dir, progress)
        if not success:
            return None, f"❌ Conversion failed: {msg}", ""

        # Create new repo
        progress(0.8, desc="Creating new repository...")
        api.create_repo(
            repo_id=new_repo_id,
            private=private_repo,
            repo_type="model",
            exist_ok=True
        )

        # Generate README
        readme = f"""---
library_name: diffusers
tags:
- gguf
- unet
- diffusion
- converted-by-gradio
---

# GGUF UNet Model

Converted from: [`{src_repo_id}`](https://huggingface.co/{src_repo_id})  
File: `unet.safetensors` → `unet.gguf`

Converted by: {user_name}  
Date: {datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')}

> ⚠️ This is a GGUF-quantized UNet for storage efficiency. Use with compatible GGUF-aware inference engines.
"""
        with open(os.path.join(output_dir, "README.md"), "w") as f:
            f.write(readme)

        # Upload
        progress(0.9, desc="Uploading to Hugging Face Hub...")
        api.upload_folder(
            repo_id=new_repo_id,
            folder_path=output_dir,
            repo_type="model",
            token=hf_token,
            commit_message="Upload UNet GGUF conversion"
        )

        progress(1.0, desc="✅ Done!")
        result_html = f"""
✅ Success!  
Your GGUF UNet is uploaded to: [{new_repo_id}](https://huggingface.co/{new_repo_id})  
Visibility: {'Private' if private_repo else 'Public'}
"""
        return gr.HTML(result_html), "✅ Conversion and upload successful!", ""

    except Exception as e:
        return None, f"❌ Error: {str(e)}", ""
    finally:
        shutil.rmtree(temp_dir, ignore_errors=True)
        shutil.rmtree(output_dir, ignore_errors=True)

# --- Gradio UI ---
with gr.Blocks(title="UNet Safetensors → GGUF Converter") as demo:
    gr.Markdown("# 🔄 UNet (Safetensors) to GGUF Converter")
    gr.Markdown("Converts `unet.safetensors` from a Hugging Face model repo to GGUF format for compact storage.")

    with gr.Row():
        with gr.Column():
            repo_url = gr.Textbox(
                label="Source Model Repository URL",
                placeholder="https://huggingface.co/Yabo/FramePainter",
                info="Must contain 'unet.safetensors'"
            )
            hf_token = gr.Textbox(
                label="Hugging Face Token",
                type="password",
                info="Write-access token from https://huggingface.co/settings/tokens"
            )
        with gr.Column():
            new_repo_id = gr.Textbox(
                label="New Repository ID",
                placeholder="your-username/framepainter-unet-gguf",
                info="Format: username/model-name"
            )
            private_repo = gr.Checkbox(label="Make Private", value=False)

    convert_btn = gr.Button("🚀 Convert & Upload", variant="primary")

    with gr.Row():
        status_output = gr.Markdown()
        repo_link_output = gr.HTML()

    convert_btn.click(
        fn=process_and_upload_unet_to_gguf,
        inputs=[repo_url, hf_token, new_repo_id, private_repo],
        outputs=[repo_link_output, status_output],
        show_progress=True
    )

    gr.Examples(
        examples=[
            ["https://huggingface.co/Yabo/FramePainter"]
        ],
        inputs=[repo_url]
    )

demo.launch()