Skip to content

Latest commit

 

History

History
67 lines (50 loc) · 2.94 KB

refactoring.rst

File metadata and controls

67 lines (50 loc) · 2.94 KB
Raw

Refactorings in this branch

SKIP (1) rename Provider.get_spatial_coverage() -> get_provided_image_size()

DONE (2) A common access pattern is:

self.dataset_cache.get_dataset(file).variables[VAR_NAME][..., :, :]

Given that Ozone has no time dim and the Albedo and Aerosol time dim is one, we might harmonise the code using explicit slice objects: * No time dim: i=(Ellipsis, slice(None), slice(None)) * Time is one: i=(0, slice(None), slice(None)) * Time is any: i=(time_index, slice(None), slice(None))

self.dataset_cache.get_dataset(file).variables[VAR_NAME][i]
    1. move aggregation code from cablab.util to gridtools
    2. rename temporal_weight() -> range_overlap_ratio()
    3. rename aggregate_images() -> aggregate_2d()
    1. move DatasetCache, NetCDFDatasetCache to ds_cache.py
    2. rename close_dataset() -> close_file
    3. rename close_all_datasets() -> close_all_files
    4. introduce abstract DatasetCache.close_dataset(dataset), override in NetCDFDatasetCache
DONE (5) a) rename get_source_time_ranges() -> compute_source_time_ranges()
  1. store result of compute_source_time_ranges in self._source_time_ranges
  2. introduce property source_time_ranges
DONE (6) a) rename cube_cli.py -> cube_gen.py
  1. move Provider stuff from cube.py to cube_gen.py
  2. rename cube.py -> cube_access.py, merge xarray stuff from branch 'issue_29'
  3. new cube_config.py

e) move CubeConfig into cube_config.py Resulting dependencies:

cube_access: cube_gen, cube_config cube_gen: cube_config any provider: cube_gen, cube_config
  1. It seems that util.Config is only used by cube_gen and provider tests! If so, move to cube_config.py
DONE (8) Provider interface:
rename get_spatial_coverage(self) -> spatial_coverage, make abstract property rename get_variable_descriptors(self) -> variable_descriptors, make abstract property

DONE (9) must address netCDF variable renamings. Introduce var attribute 'source_name'

DONE (10) Replace special compute_variable_images_from_sources() impl. by generic one in base class.

  1. Review all variable_descriptors() impls (Fabian and Gunnar): * check: target variable names, use CF names or just CF-compliant names? * check: why are the different 'fill_value' values? we should use a common one for each data type, e.g. np.nan * check: find correct 'standard_name' for each variable, current ones are mostly wrong * check: 'long_name' name is often wrong or confusing,

    e.g. 'level 3b fractional snow cover (%) aggregated monthly' when our period is 8 days. All lower case?

    • check: possibly add 'comment' or 'description' or 'history' and tell where data comes from

  2. Generated netCDF files and variables still lack important CF variables: * 'Conventions' * 'history' * 'coordinates' * 'lat_bnd', 'lon_bnds' + their 'standard_name' values, see #20, #30