"""Dataset mask test."""

from affine import Affine
import numpy as np
import pytest
from affine import Affine

import rasterio
from rasterio.enums import Resampling
from rasterio.errors import NodataShadowWarning
from rasterio.crs import CRS


# Setup test arrays
red = np.array([[0, 0, 0],
                [0, 1, 1],
                [1, 0, 1]]).astype('uint8') * 255

grn = np.array([[0, 0, 0],
                [1, 0, 1],
                [1, 0, 1]]).astype('uint8') * 255

blu = np.array([[0, 0, 0],
                [1, 1, 0],
                [1, 0, 1]]).astype('uint8') * 255

# equivalent to alp = red | grn | blu
# valid data anywhere there is at least one R, G or B value
alp = np.array([[0, 0, 0],
                [1, 1, 1],
                [1, 0, 1]]).astype('uint8') * 255

# mask might be constructed using different tools
# and differ from a strict interpretation of rgb values
msk = np.array([[0, 0, 0],
                [1, 1, 1],
                [1, 1, 1]]).astype('uint8') * 255

alldata = np.array([[1, 1, 1],
                    [1, 1, 1],
                    [1, 1, 1]]).astype('uint8') * 255

# boundless window ((1, 4, (1, 4))
alp_shift_lr = np.array([[1, 1, 0],
                         [0, 1, 0],
                         [0, 0, 0]]).astype('uint8') * 255

# whole mask resampled to (1, 5, 5) array
resampmask = np.array([[0, 0, 0, 0, 0],
                       [0, 0, 0, 0, 0],
                       [1, 1, 1, 1, 1],
                       [1, 1, 0, 1, 1],
                       [1, 1, 0, 1, 1]]).astype('uint8') * 255

# whole mask resampled to (1, 5, 5) array
resampave = np.array([[0, 0, 0, 0, 0],
                      [1, 1, 1, 1, 1],
                      [1, 1, 1, 1, 1],
                      [1, 1, 1, 1, 1],
                      [1, 1, 0, 1, 1]]).astype('uint8') * 255


@pytest.fixture(scope='function')
def tiffs(tmpdir):

    _profile = {
        'transform': Affine(5.0, 0.0, 0.0, 0.0, -5.0, 0.0),
        'crs': CRS({'init': 'epsg:4326'}),
        'driver': 'GTiff',
        'dtype': 'uint8',
        'height': 3,
        'width': 3}

    # 1. RGB without nodata value
    prof = _profile.copy()
    prof['count'] = 3
    prof['nodata'] = None
    with rasterio.open(str(tmpdir.join('rgb_no_ndv.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)

    # 2. RGB with nodata value
    prof = _profile.copy()
    prof['count'] = 3
    prof['nodata'] = 0
    with rasterio.open(str(tmpdir.join('rgb_ndv.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)

    # 3. RGBA without nodata value
    prof = _profile.copy()
    prof['count'] = 4
    prof['nodata'] = None
    with rasterio.open(str(tmpdir.join('rgba_no_ndv.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)
        dst.write(alp, 4)

    # 4. RGBA with nodata value
    prof = _profile.copy()
    prof['count'] = 4
    prof['nodata'] = 0
    with rasterio.open(str(tmpdir.join('rgba_ndv.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)
        dst.write(alp, 4)

    # 5. RGB with msk
    prof = _profile.copy()
    prof['count'] = 3
    with rasterio.open(str(tmpdir.join('rgb_msk.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)
        dst.write_mask(msk)

    # 6. RGB with msk (internal)
    prof = _profile.copy()
    prof['count'] = 3
    with rasterio.Env(GDAL_TIFF_INTERNAL_MASK=True):
        with rasterio.open(str(tmpdir.join('rgb_msk_internal.tif')),
                           'w', **prof) as dst:
            dst.write(red, 1)
            dst.write(grn, 2)
            dst.write(blu, 3)
            dst.write_mask(msk)

    # 7. RGBA with msk
    prof = _profile.copy()
    prof['count'] = 4
    with rasterio.open(str(tmpdir.join('rgba_msk.tif')), 'w', **prof) as dst:
        dst.write(red, 1)
        dst.write(grn, 2)
        dst.write(blu, 3)
        dst.write(alp, 4)
        dst.write_mask(msk)

    return tmpdir


def test_no_ndv(tiffs):
    with rasterio.open(str(tiffs.join('rgb_no_ndv.tif'))) as src:
        assert np.array_equal(src.dataset_mask(), alldata)


def test_rgb_ndv(tiffs):
    with rasterio.open(str(tiffs.join('rgb_ndv.tif'))) as src:
        res = src.dataset_mask()
        assert res.dtype.name == "uint8"
        assert np.array_equal(src.dataset_mask(), alp)


def test_rgba_no_ndv(tiffs):
    with rasterio.open(str(tiffs.join('rgba_no_ndv.tif'))) as src:
        assert np.array_equal(src.dataset_mask(), alp)


def test_rgba_ndv(tiffs):
    with rasterio.open(str(tiffs.join('rgba_ndv.tif'))) as src:
        with pytest.warns(NodataShadowWarning):
            res = src.dataset_mask()
        assert np.array_equal(res, alp)


def test_rgb_msk(tiffs):
    with rasterio.open(str(tiffs.join('rgb_msk.tif'))) as src:
        assert np.array_equal(src.dataset_mask(), msk)
        # each band's mask is also equal
        for bmask in src.read_masks():
            assert np.array_equal(bmask, msk)


def test_rgb_msk_int(tiffs):
    with rasterio.open(str(tiffs.join('rgb_msk_internal.tif'))) as src:
        assert np.array_equal(src.dataset_mask(), msk)


def test_rgba_msk(tiffs):
    with rasterio.open(str(tiffs.join('rgba_msk.tif'))) as src:
        # mask takes precedent over alpha
        assert np.array_equal(src.dataset_mask(), msk)


@pytest.mark.parametrize("kwds,expected", [(dict(window=((1, 4), (1, 4)), boundless=True), alp_shift_lr), (dict(out_shape=(1, 5, 5)), resampmask), (dict(out=np.zeros((1, 5, 5), dtype=np.uint8)), resampmask)])
def test_kwargs(tiffs, kwds, expected):
    with rasterio.open(str(tiffs.join('rgb_ndv.tif'))) as src:
        result = src.dataset_mask(**kwds)
        assert np.array_equal(expected, result)


def test_indexes_not_supported(tiffs):
    with rasterio.open(str(tiffs.join('rgb_ndv.tif'))) as src:
        with pytest.raises(TypeError):
            src.dataset_mask(indexes=1)


def test_kwargs_resampling(tiffs):
    with rasterio.open(str(tiffs.join('rgb_ndv.tif'))) as src:
        other = src.dataset_mask(out_shape=(1, 5, 5), resampling=Resampling.bilinear) != 0
        other = other.astype(np.uint8) * 255
        assert np.array_equal(resampave, other)