Add helper method for checking areas in compositors #224
Conversation
satpy/composites/__init__.py
Outdated
| @@ -44,6 +44,7 @@ | |||
| from satpy.utils import sunzen_corr_cos, atmospheric_path_length_correction | |||
| from satpy.writers import get_enhanced_image | |||
| from satpy import CHUNK_SIZE | |||
| from pyresample.geometry import AreaDefinition | |||
There was a problem hiding this comment.
F401 'pyresample.geometry.AreaDefinition' imported but unused
There was a problem hiding this comment.
geeze, it's making comments too
|
I'm now leaning towards this always raising incompatible areas and never correcting. |
| "mapped to the same area as the " | ||
| "low resolution bands. Must " | ||
| "resample first.") | ||
| datasets = self.check_areas(datasets + optional_datasets) |
There was a problem hiding this comment.
some compositors (palettecompositor) have a palette as optional dependency, and the palette has no area attribute, and has a different shape.
There was a problem hiding this comment.
So maybe check that the dimensions are called x and y before going further ? (although I know we're not there yet)
There was a problem hiding this comment.
Checking dimensions sounds good and makes sense. However this specific line is just for ratio sharpening so it should never be a palette.
| @@ -382,6 +435,7 @@ def __call__(self, projectables, optional_datasets=None, **info): | |||
| sunalt, suna = get_alt_az(vis.attrs['start_time'], lons, lats) | |||
| suna = xu.rad2deg(suna) | |||
| sunz = sun_zenith_angle(vis.attrs['start_time'], lons, lats) | |||
| # FIXME: Make it daskified | |||
There was a problem hiding this comment.
an eays fix is to compute lons and lats
satpy/composites/__init__.py
Outdated
| if len(data_arrays) == 1: | ||
| return data_arrays | ||
|
|
||
| if not all(x.shape == data_arrays[0].shape for x in data_arrays[1:]): |
There was a problem hiding this comment.
shouldn't we be checking only the x and y dimension sizes ? and what guarantees that all data array have the same number of dimensions ?
There was a problem hiding this comment.
Good point. This was done assuming we were dealing with 2D reader datasets. Maybe I'm not completely xarray-minded yet.
satpy/composites/__init__.py
Outdated
| @@ -44,6 +44,7 @@ | |||
| from satpy.utils import sunzen_corr_cos, atmospheric_path_length_correction | |||
| from satpy.writers import get_enhanced_image | |||
| from satpy import CHUNK_SIZE | |||
| from pyresample.geometry import AreaDefinition | |||
There was a problem hiding this comment.
F401 'pyresample.geometry.AreaDefinition' imported but unused
|
|
||
| def test_mult_ds_diff_dims(self): | ||
| """Test that datasets with different dimensions still pass.""" | ||
| from satpy.composites import CompositeBase, IncompatibleAreas |
There was a problem hiding this comment.
F401 'satpy.composites.IncompatibleAreas' imported but unused
djhoese
changed the title
|
Hm maybe this one isn't ready. I get an unexpected error about the time coordinates when making the simulated green band for ABI. |
I made this PR early so that @mraspaud and I and others could discuss this because it may be controversial and I wanted to figure this out before I spent time implementing everything.
Right now this PRs solution for datasets with equal resolutions but different coordinates is to just remove the coordinates. @mraspaud voted to have it just choose the first datasets coordinates. I had already implemented what is in this PR currently so I didn't do this yet. My argument is that this may be a bad idea because coordinates for multiple composites may not be the same depending on what the input datasets were.
git diff origin/develop **/*py | flake8 --diff