src/utils.py

"""Common helpers: config loading with ${var} interpolation, seeding, IO."""
from __future__ import annotations

import json
import os
import random
import re
from pathlib import Path
from typing import Any, Dict

import numpy as np
import torch
import yaml

_INTERP_RE = re.compile(r"\$\{([^}]+)\}")


def _resolve(value: Any, root: Dict[str, Any]) -> Any:
    if isinstance(value, str):
        prev = None
        while prev != value and "${" in value:
            prev = value

            def repl(m: re.Match) -> str:
                key = m.group(1)
                node: Any = root
                for part in key.split("."):
                    node = node[part]
                return str(node)

            value = _INTERP_RE.sub(repl, value)
        return value
    if isinstance(value, dict):
        return {k: _resolve(v, root) for k, v in value.items()}
    if isinstance(value, list):
        return [_resolve(v, root) for v in value]
    return value


def load_config(path: str | Path) -> Dict[str, Any]:
    # encoding is explicit: several configs carry an em-dash in their header
    # comment, and open() defaults to the locale encoding, which is ASCII here.
    # Without this, deberta_toxigen/newsgroups/yelp fail at byte 29 with a
    # UnicodeDecodeError that names the YAML reader rather than the file.
    with open(path, "r", encoding="utf-8") as f:
        cfg = yaml.safe_load(f)
    cfg = _resolve(cfg, cfg)
    return cfg


def seed_everything(seed: int) -> None:
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    if torch.cuda.is_available():
        torch.cuda.manual_seed_all(seed)


def ensure_dir(path: str | Path) -> Path:
    p = Path(path)
    p.mkdir(parents=True, exist_ok=True)
    return p


def save_json(obj: Any, path: str | Path) -> None:
    with open(path, "w") as f:
        json.dump(obj, f, indent=2, default=str)


def load_json(path: str | Path) -> Any:
    with open(path, "r") as f:
        return json.load(f)


def device_from_cfg(cfg: Dict[str, Any]) -> torch.device:
    requested = cfg.get("inference", {}).get("device", "cuda")
    if requested == "cuda" and not torch.cuda.is_available():
        return torch.device("cpu")
    return torch.device(requested)