v1.44.5

1.44.5 (2025-01-10)

• ENH: Add the ability to round rasters values in ci.assert_raster_max_mismatch
• FIX: Fix regression in rasters.collocate where odc.geo.xr_reproject was not used correctly.
• FIX: Fix when trying to pop non-existing nodata keyword in rasters.collocate

v1.44.4

1.44.4 (2025-01-07)

• FIX: Fix regression in rasters.any_raster_to_xr_ds (right Exception not thrown, UnboundLocalError thrown instead when trying to return non-existing variable)

v1.44.3

1.44.3 (2025-01-06)

• FIX: Manage case where we have a pd.Dataframe instead of a gpd.GeoDataFrame in vectors.read (reading a .dbf file for instance)
• FIX: Simplify decorator function of rasters.read, to better check the input types and to have a clearer function name and clearer exceptions
• FIX: Simplify decorators rasters.any_raster_to_xr_ds and rasters_rio.any_raster_to_rio_ds to better check the input types and to have clearer exceptions

v1.44.2

1.44.2 (2024-12-23)

• ENH: Drop isort, black and flake8 and use ruff
• ENH: Use pyproject.toml instead of setup.py
• FIX: Fix too broad exception in case of geopandas >= 1.0
• FIX: Fix deprecation warning for get_nodata_value_from_dtype in rasters_rio
• FIX: Force blocksize to 128 when writing small COGs on disk (in order to have multiple overview levels)
• FIX: Use np.tan in rasters.slope
• FIX: Allow str as paths in ci.assert_files_equal
• FIX: Better alignement between rasters.read function and rasters.any_raster_to_xr_ds decorator
• FIX: Fix rasters.sieve function with xr.apply_ufunc
• OPTIM: Compute the spatial index by default in vectors.read (set vectors.read(..., compute_sindex=False) if you don't want to compute them)
• CI: Rename CI folder and remove unnecessary intermediate folder

v1.44.1

1.44.1 (2024-12-12)

• OPTIM: Don't download an archive stored on the cloud when trying to read a vector stored inside it in vectors.read
• OPTIM: Don't download files stored on cloud when applying ci.assert_files_equal on them
• OPTIM: Offer the ability to give the archived file list directly to path.get_archived_file_list and files.read_archived_file, as this operation is expensive when done with large archives stored on the cloud (and thus better done only once).
  Propagated into path.get_archived_path, path.get_archived_rio_path, vectors.read, xml.read_archive, files.read_archived_xml and files.read_archived_html
• CI: Test both Dask's multithreaded and local cluster in CI
• DEPR: Deprecate xml_regex and file_regex arguments from above-mentioned functions in favor of a harmonized regex argument.
• DEPS: Reduce dask-related dependencies to the libraries related to this project (abandon dask[complete] as we don't need most of the embedded libraries).
• DOCS: Add notebooks to document best practises

v1.44.1.dev1

• OPTIM: Offer the ability to give the archived file list directly to path.get_archived_file_list and files.read_archived_file, as this operation is expensive when done with large archives stored on the cloud (and thus better done only once).
  Propagated into path.get_archived_path, path.get_archived_rio_path, vectors.read, files.read_archived_xml and files.read_archived_html

v1.44.1.dev0

1.44.1.dev0 (2024-12-09)

• OPTIM: Don't download an archive stored on the cloud when trying to read a vector stored inside it in vectors.read
• OPTIM: Don't download files stored on cloud when applying ci.assert_files_equal on them

v1.44.0

1.44.0 (2024-12-09)

• BREAKING CHANGE: Renaming of rasters(_rio).write argument path to output_path to avoid shadowing sertit.path module. Older argument is deprecated.
• ENH: Use odc.geo.xr.mask instead of rasterio.mask to be dask-compatible in rasters (true for paint and mask) (#27)
• ENH: Use odc.geo.xr.xr_reproject instead of rioxarray.reproject_match to be dask-compatible in rasters.collocate (#27)
• ENH: Use xarray-spatial to be dask-compatible in rasters.slope and rasters.hillshade. (#27)
• ENH: Add support for aspect function (only in rasters). (#29)
• FIX: Fix the ability to save COGs with any dtype with Dask, with the workaround described here (don't compute statistics for problematic dtypes)
• FIX: Better separability of dask (it has its own module now): don't create a client if the user doesn't specify it (as it is not required anymore in Lock). This should remove the force-use of dask.
• OPTIM: For arrays with same shape and CRS, replace only the coordinates of other by ref's in rasters.collocate
• OPTIM: Call dask.optimize before any dask computation (#27)
• FIX: Fix vectorization with dask arrays (and remove the silent failure in case of exception when computing) (#27)
• DEPS: Add an optional dependency to xarray-spatial for daskified surface tools, such as hillshade and slope

⚠️ We selected for now xarray-spatial dependency for DEM-related functiions, but this may not be the final selection.
When a bit more mature, it is rather likely that geoutils and xdem will be selected instead.
See this issue for dask-related choices.

v1.44.0.rc1

1.44.0.rc1 (2024-12-09)

• BREAKING CHANGE: Renaming of rasters(_rio).write argument path to output_path to avoid shadowing sertit.path module. Older argument is deprecated.
• ENH: Use odc.geo.xr.mask instead of rasterio.mask to be dask-compatible in rasters (true for paint and mask) (#27)
• ENH: Use odc.geo.xr.xr_reproject instead of rioxarray.reproject_match to be dask-compatible in rasters.collocate (#27)
• ENH: Use xarray-spatial to be dask-compatible in rasters.slope and rasters.hillshade. (#27)
• ENH: Add support for aspect function (only in rasters). (#29)
• FIX: Fix the ability to save COGs with any dtype with Dask, with the workaround described here (don't compute statistics for problematic dtypes)
• FIX: Better separability of dask (it has its own module now): don't create a client if the user doesn't specify it (as it is not required anymore in Lock). This should remove the force-use of dask.
• OPTIM: For arrays with same shape and CRS, replace only the coordinates of other by ref's in rasters.collocate
• OPTIM: Call dask.optimize before any dask computation (#27)
• FIX: Fix vectorization with dask arrays (and remove the silent failure in case of exception when computing) (#27)
• DEPS: Add an optional dependency to xarray-spatial for daskified surface tools, such as hillshade and slope

⚠️ We selected for now xarray-spatial dependency for DEM-related functiions, but this may not be the final selection.
When a bit more mature, it is rather likely that geoutils and xdem will be selected instead.
See this issue for dask-related choices.

v1.43.4

1.43.4 (2024-11-28)

• FIX: Fix regression in files.read_json due to Python 3.11 where datatime.fromsioformat parses more than the output of isoformat()