bioai/paths.py

"""
Centralized path resolution for bioai.

All runtime paths are derived from the project root, which is the directory
containing the ``bioai/`` package. This makes the codebase portable across
machines (no hardcoded ``/home/z/my-project`` paths).

Resolution order:
  1. Walk up from this file (``bioai/paths.py``) until we find a directory
     that contains a ``bioai`` subdirectory. That's the project root.
  2. If that fails (e.g., we're inside a frozen binary), fall back to the
     current working directory.

Override the project root by setting the ``BIOAI_PROJECT_ROOT`` env var
(useful for Docker, where WORKDIR=/app is the project root).
"""

from __future__ import annotations

import os
from pathlib import Path


def _find_project_root() -> Path:
    # Allow explicit override
    env = os.environ.get("BIOAI_PROJECT_ROOT")
    if env:
        p = Path(env).expanduser().resolve()
        if p.is_dir():
            return p

    # Walk up from this file until we find a directory containing bioai/
    here = Path(__file__).resolve().parent  # .../bioai/
    for candidate in [here, *here.parents]:
        if (candidate / "bioai").is_dir():
            return candidate

    # Last resort: current working directory
    return Path.cwd()


PROJECT_ROOT: Path = _find_project_root()

# Standard subdirectories
DATA_DIR: Path = PROJECT_ROOT / "data"
SYNTHETIC_DIR: Path = DATA_DIR / "synthetic"
EXTERNAL_DIR: Path = DATA_DIR / "external"
PROCESSED_DIR: Path = DATA_DIR / "processed"
CHECKPOINTS_DIR: Path = PROJECT_ROOT / "checkpoints"
CACHE_DIR: Path = PROJECT_ROOT / ".fireworks_cache"

# Standard file paths
DEFAULT_PEST_FASTA: Path = SYNTHETIC_DIR / "pest_transcripts.fasta"
DEFAULT_SAFETY_FASTA: Path = SYNTHETIC_DIR / "safety_transcripts.fasta"
DEFAULT_TRAINING_CSV: Path = PROCESSED_DIR / "training_data.csv"
SIRNA_CHECKPOINT: Path = CHECKPOINTS_DIR / "sirna_best.pt"
VAE_CHECKPOINT: Path = CHECKPOINTS_DIR / "vae_best.pt"
PINN_CHECKPOINT: Path = CHECKPOINTS_DIR / "pinn_best.pt"


def ensure_dirs() -> None:
    """Create the standard subdirectories if they don't exist.

    Called at module import so any subsequent path is safe to write to.
    """
    for d in (DATA_DIR, SYNTHETIC_DIR, EXTERNAL_DIR, PROCESSED_DIR, CHECKPOINTS_DIR, CACHE_DIR):
        d.mkdir(parents=True, exist_ok=True)


# Run once on import
ensure_dirs()