bootstrap.py

#!/usr/bin/env python3
"""
Bootstrap module - Downloads ver20 from private Space.
  mbok_dev / mbok-dev  → DLPO/habadashi_dev
  mbok (その他)        → DLPO/habadashi
"""

import os
from pathlib import Path
from huggingface_hub import snapshot_download

def _resolve_repo_id() -> str:
    """SPACE_ID 環境変数から参照先 private Space の repo_id を決定する。"""
    space_id = os.environ.get("SPACE_ID", "").lower()
    if "mbok_dev" in space_id or "mbok-dev" in space_id:
        return "DLPO/habadashi_dev"
    return "DLPO/habadashi"

def get_hf_token():
    """Get HF_TOKEN from environment variable"""
    token = os.environ.get("HF_TOKEN")
    if not token:
        print("[BOOTSTRAP_ERROR] HF_TOKEN not found in environment")
        raise ValueError(
            "HF_TOKEN not found. Please set HF_TOKEN environment variable "
            "in HF Space Secrets to access private repository."
        )
    print(f"[BOOTSTRAP] HF_TOKEN found (length: {len(token)}, 末尾3文字: ...{token[-3:]})")
    return token

def download_private_app(force_download: bool = False):
    """
    Download ver20 application from private Space.
    参照先は SPACE_ID 環境変数で決定:
      mbok_dev / mbok-dev → DLPO/habadashi_dev
      mbok (その他)       → DLPO/habadashi

    Args:
        force_download: If True, re-download even if already cached

    Returns:
        Path: Local directory containing downloaded ver20 files
    """
    repo_id = _resolve_repo_id()
    repo_type = "space"
    local_dir = Path("./private_app")
    
    print(f"[BOOTSTRAP] Starting download")
    print(f"[BOOTSTRAP] repo_id={repo_id} repo_type={repo_type}")
    print(f"[BOOTSTRAP] local_dir={local_dir} force_download={force_download}")
    print(f"[BOOTSTRAP] local_dir.exists()={local_dir.exists()}")
    
    # Check if already downloaded (skip if exists and not forced)
    if local_dir.exists() and not force_download:
        # Verify essential files exist
        app_py = local_dir / "app.py"
        print(f"[BOOTSTRAP] Checking cache: app.py exists={app_py.exists()}")
        if app_py.exists():
            print(f"[BOOTSTRAP] Using cached download: {local_dir}")
            # Verify structure
            _verify_downloaded_structure(local_dir)
            return local_dir
        else:
            print(f"[BOOTSTRAP] Cache invalid (app.py missing), will re-download")
    
    print(f"[BOOTSTRAP] Starting fresh download from {repo_id}...")
    
    try:
        token = get_hf_token()
        
        # Download entire Space repository (removed local_dir_use_symlinks - deprecated)
        print(f"[BOOTSTRAP] Calling snapshot_download...")
        snapshot_download(
            repo_id=repo_id,
            repo_type=repo_type,
            local_dir=str(local_dir),
            token=token,
        )
        
        print(f"[BOOTSTRAP] Download complete: {local_dir}")
        
        # Verify essential structure
        _verify_downloaded_structure(local_dir)
        
        return local_dir
        
    except Exception as e:
        print(f"[BOOTSTRAP_ERROR] Download failed: {e}")
        import traceback
        print(f"[TRACEBACK]\n{traceback.format_exc()}")
        raise

def _verify_downloaded_structure(local_dir: Path):
    """Verify downloaded structure has essential components"""
    print(f"[BOOTSTRAP_VERIFY] Checking downloaded structure...")
    
    essential_files = [
        "app.py",
        "requirements.txt",
    ]
    
    essential_dirs = [
        "lib",
        "exec", 
        "core",
        "presentation",
    ]
    
    for file_name in essential_files:
        file_path = local_dir / file_name
        exists = file_path.exists()
        print(f"[BOOTSTRAP_VERIFY] {file_name}: {'EXISTS' if exists else 'MISSING'}")
        if not exists and file_name == "app.py":
            raise FileNotFoundError(
                f"Critical file missing: {file_name}. "
                f"Expected at: {file_path}"
            )
    
    for dir_name in essential_dirs:
        dir_path = local_dir / dir_name
        exists = dir_path.exists() and dir_path.is_dir()
        print(f"[BOOTSTRAP_VERIFY] {dir_name}/: {'EXISTS' if exists else 'MISSING'}")
    
    print(f"[BOOTSTRAP_VERIFY] Verification complete")

if __name__ == "__main__":
    # Test download
    download_private_app()
    print("🎉 Bootstrap test successful!")