utils/paths.py

# utils/paths.py
from pathlib import Path
import os, re, uuid, shutil

APP_ROOT = Path(__file__).resolve().parents[1]
DATA_ROOT = APP_ROOT / "data"
TMP_ROOT  = APP_ROOT / "tmp"
for p in (DATA_ROOT, TMP_ROOT, APP_ROOT / ".hf", APP_ROOT / ".torch"):
    p.mkdir(parents=True, exist_ok=True)

os.environ.setdefault("HF_HOME", str(APP_ROOT / ".hf"))
os.environ.setdefault("TORCH_HOME", str(APP_ROOT / ".torch"))

def safe_name(name: str, default="file"):
    base = re.sub(r"[^A-Za-z0-9._-]+", "_", (name or default))
    return (base or default)[:120]

def job_dir(prefix="job"):
    d = DATA_ROOT / f"{prefix}-{uuid.uuid4().hex[:8]}"
    d.mkdir(parents=True, exist_ok=True)
    return d

def disk_stats(p: Path = APP_ROOT) -> str:
    try:
        total, used, free = shutil.disk_usage(str(p))
        mb = lambda x: x // (1024 * 1024)
        return f"disk(total={mb(total)}MB, used={mb(used)}MB, free={mb(free)}MB)"
    except Exception:
        return "disk(n/a)"