backend/utils/runtime_paths.py

"""Keep temp, HF cache, and scratch I/O on the configured DATA_DIR volume.

Windows returns ``[Errno 22] Invalid argument`` when the system temp drive (often
C:) is nearly full while FAISS / upload / embedding caches still write there.
"""

from __future__ import annotations

import os
from pathlib import Path

from backend.config import settings


def ensure_data_drive_runtime_dirs() -> Path:
    """Point TMP/TEMP and HF caches at ``DATA_DIR`` (safe on D: etc.)."""
    root = settings.data_dir_path
    tmp = root / "tmp"
    tmp.mkdir(parents=True, exist_ok=True)
    os.environ["TMP"] = str(tmp)
    os.environ["TEMP"] = str(tmp)

    hf = root / "hf_cache"
    hf.mkdir(parents=True, exist_ok=True)
    os.environ.setdefault("HF_HOME", str(hf))
    os.environ.setdefault("SENTENCE_TRANSFORMERS_HOME", str(hf / "sentence_transformers"))
    os.environ.setdefault("TRANSFORMERS_CACHE", str(hf / "transformers"))
    os.environ.setdefault("TORCH_HOME", str(hf / "torch"))
    return tmp