"""Config loader + light schema validation (metadata-independent).

Loads the project YAML configs and checks that required top-level keys are
present, so a typo fails loudly rather than silently. Does NOT encode any ECMWF
GRIB assumptions.
"""
import os

_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", ".."))
CONFIG_DIR = os.path.join(_ROOT, "config")

# minimal required top-level keys per config file (extend as configs grow)
REQUIRED = {
    "targets": ["period", "horizon", "p90", "ehd_threshold", "weeks"],
    "domains": ["target_grid", "region_mask", "predictor_domains"],
    "data": ["observations", "ecmwf"],
    "verification": ["headline_weeks", "metrics", "leakage_rule", "regions"],
    "variables": ["surface", "pressure"],
    "model": ["architecture", "outputs"],
    "training": ["cross_validation", "compute_environment"],
}


def load_config(name):
    """Load config/<name>.yaml as a dict and validate required keys."""
    try:
        import yaml
    except Exception as e:  # pragma: no cover
        raise RuntimeError("pyyaml not available; install it or parse manually") from e
    path = os.path.join(CONFIG_DIR, f"{name}.yaml")
    if not os.path.exists(path):
        raise FileNotFoundError(path)
    with open(path) as f:
        cfg = yaml.safe_load(f)
    missing = [k for k in REQUIRED.get(name, []) if k not in (cfg or {})]
    if missing:
        raise KeyError(f"{name}.yaml missing required keys: {missing}")
    return cfg


def load_all():
    return {name: load_config(name) for name in REQUIRED}