"""VnGAD gridded-Tx observation source (metadata-independent; no GRIB).

Depends only on the observational Tx NetCDF, so it is safe to build now and is
what target construction and (later) forecast-to-observation alignment read
from. Originals are opened read-only and never modified.
"""
import os

from .interfaces import ObservationSource

_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", ".."))
DEFAULT_TX = os.path.join(_ROOT, "data", "VnGAD_daily-0.25_1961-2023_tasmax.nc")


class VnGADObservations(ObservationSource):
    """Local (UTC+7) daily-max Tx on the 0.25 deg VnGAD grid."""

    def __init__(self, path=DEFAULT_TX, var="tasmax"):
        import xarray as xr
        self.path = path
        self.var = var
        self._ds = xr.open_dataset(path)   # lazy; values read on demand
        self.lat = self._ds["lat"].values
        self.lon = self._ds["lon"].values

    def dates(self):
        """Observation local-day axis as datetime64[D]."""
        return self._ds["time"].values.astype("datetime64[D]")

    def tx(self, start=None, end=None):
        """Tx (time, lat, lon) in degC for [start, end] (inclusive, by date)."""
        da = self._ds[self.var]
        if start is not None or end is not None:
            da = da.sel(time=slice(start, end))
        return da

    def valid_mask(self, ref_years=(2003, 2022)):
        """Time-stable valid-data mask (lat, lon) over the reference period,
        built from ACTUAL valid data (not a hard-coded land mask)."""
        import numpy as np
        d = self.dates()
        yr = d.astype("datetime64[Y]").astype(int) + 1970
        sel = (yr >= ref_years[0]) & (yr <= ref_years[1])
        arr = self._ds[self.var].values[sel]
        return np.isfinite(arr).any(0)