def as_geo_docs(self) -> Tuple[CRS, Dict[str, GridDoc], Dict[str, MeasurementDoc]]:
"""Calculate combined geo information for metadata docs"""
if not self._measurements_per_grid:
return None, None, None
grid_docs: Dict[str, GridDoc] = {}
measurement_docs: Dict[str, MeasurementDoc] = {}
crs = None
for grid_name, (grid, measurements) in self._as_named_grids().items():
# Validate assumption: All grids should have same CRS
if crs is None:
crs = grid.crs
# TODO: CRS equality is tricky. This may not work.
# We're assuming a group of measurements specify their CRS
# the same way if they are the same.
elif grid.crs != crs:
raise ValueError(
f"Measurements have different CRSes in the same dataset:\n"
f"\t{crs.to_string()!r}\n"
f"\t{grid.crs.to_string()!r}\n"
)
grid_docs[grid_name] = GridDoc(grid.shape, grid.transform)
for measurement_name, measurement_path in measurements.items():
# No measurement groups in the doc: we replace with underscores.
measurement_name = measurement_name.replace(":", "_")
measurement_docs[measurement_name] = MeasurementDoc(
path=measurement_path.path,
layer=measurement_path.layer,
grid=grid_name if grid_name != "default" else None,
)
return crs, grid_docs, measurement_docs
def _band_to_measurement(band: Dict, dataset_location: str) -> MeasurementDoc:
"""Create EO3 measurement from an EO1 band dict"""
return MeasurementDoc(
path=band.get("path"),
band=band.get("band"),
layer=band.get("layer"),
name=band.get("name"),
alias=band.get("label"),
)
def as_geo_docs(
self) -> Tuple[CRS, Dict[str, GridDoc], Dict[str, MeasurementDoc]]:
"""Calculate combined geo information for metadata docs"""
if not self._measurements_per_grid:
return None, None, None
# Order grids from most to fewest measurements.
# PyCharm's typing seems to get confused by the sorted() call.
# noinspection PyTypeChecker
grids_by_frequency: List[Tuple[GridSpec,
Dict[str, MeasPath]]] = sorted(
self._measurements_per_grid.items(),
key=lambda k: len(k[1]),
reverse=True)
grid_docs: Dict[str, GridDoc] = {}
measurement_docs: Dict[str, MeasurementDoc] = {}
crs = grids_by_frequency[0][0].crs
for i, (grid, measurements) in enumerate(grids_by_frequency):
# TODO: CRS equality is tricky. This may not work.
# We're assuming a group of measurements specify their CRS
# the same way if they are the same.
if grid.crs != crs:
raise ValueError(
f"Measurements have different CRSes in the same dataset:\n"
f"\t{crs.to_string()!r}\n"
f"\t{grid.crs.to_string()!r}\n")
# The grid with the most measurements.
if i == 0:
grid_name = "default"
else:
grid_name = _find_a_common_name(list(measurements.keys()))
# If another grid already has this name: TODO: make both grid names more specific?
if grid_name in grid_docs:
raise NotImplementedError(
f"Clashing grid names. Needs a recalculation. "
f"Name {grid_name!r}, but have {tuple(grid_docs.keys())!r}"
)
# There was no common prefix. Just concat all band names.
# Perhaps we just fallback to enumeration in these weird cases. grid a, grid b etc....
if not grid_name:
grid_name = "_".join(measurements.keys())
grid_docs[grid_name] = GridDoc(grid.shape, grid.transform)
for measurement_name, measurement_path in measurements.items():
# No measurement groups in the doc: we replace with underscores.
measurement_name = measurement_name.replace(":", "_")
measurement_docs[measurement_name] = MeasurementDoc(
path=measurement_path.path,
layer=measurement_path.layer,
grid=grid_name if grid_name != "default" else None,
)
return crs, grid_docs, measurement_docs
def test_esri_prepare():
tif_url = ESRI_ANTIMERIDIAN_OVERLAP_TIF.resolve().as_uri()
dataset_doc = as_eo3(tif_url)
# Varies
dataset_doc.properties.pop("odc:processing_datetime")
assert_expected_eo3(
DatasetDoc(
id=UUID("411f3641-bc4a-501d-99e5-6c5c47379715"),
product=ProductDoc("esri_land_cover"),
accessories={},
lineage={},
label="esri_land_cover_60K_2020-01-01",
locations=[
tif_url,
],
properties=Eo3Dict(
{
"datetime": datetime(2020, 1, 1, 0, 0, tzinfo=timezone.utc),
"dtr:end_datetime": datetime(2021, 1, 1, 0, 0, tzinfo=timezone.utc),
"dtr:start_datetime": datetime(
2020, 1, 1, 0, 0, tzinfo=timezone.utc
),
"odc:product": "esri_land_cover",
"odc:region_code": "60K",
}
),
crs="epsg:32760",
measurements=dict(
classification=MeasurementDoc(
path="60K_20200101-20210101.tif",
)
),
geometry=shape(
{
"coordinates": (
(
(841600.2521048393, 7864419.112077935),
(841600.2521048393, 8221119.112077935),
(476100.25210483925, 8221119.112077935),
(476100.25210483925, 7864419.112077935),
(841600.2521048393, 7864419.112077935),
),
),
"type": "Polygon",
}
),
grids={
"default": GridDoc(
(128, 128),
Affine(
2855.46875,
0.0,
476100.25210483925,
0.0,
-2786.71875,
8221119.112077935,
),
)
},
),
dataset_doc,
)