Changelog

v0.5.1 (2025-10-02)

Contributors: Trevor James Smith (@Zeitsperre).

Fixes

• Author and license information have been updated in the top-most __init__.py, .zenodo.json, and pyproject.toml files.

Internal changes

• Removed the obsolete MANIFEST.in file.

• New pre-commit checks have been added to the .pre-commit-config.yaml file.

v0.5.0 (2025-05-05)

Bug fixes

• Scalar attributes that are not strings are no longer wrapped in tuples of length==1 (by @bzah).

Breaking changes

• Nested group handling (by @bzah):

  • Before this version, all groups were read, but conflicting variable names in-between groups would shadow data. Now, similarly to xarray.open_dataset, open_ncml accepts an optional group argument to specify which group should be read. When group is not specified, it defaults to the root group. Additionally group can be set to '*' so that every group is read and the hierarchy is flattened. In the event of conflicting variable/dimension names across groups, the conflicting name will be modified by appending '__n' where n is incremented.

• Enums are no longer transformed into CF flag_values and flag_meanings attributes, instead they are stored in the encoding["dtype"].metadata of their respective variable. This is aligned with what was done on xarray v2024.01.0 (by @bzah).

• xncml.generated.ncml_2_2 has been refactored to no longer be exposed to the package API. Users should now use objects imported directly from xncml.generated (by @Zeitsperre).

• Add support for Python 3.13 and drop support for Python 3.9 (by @Zeitsperre).

Internal changes

• Added support for running pytest with pytest-cov (by @Zeitsperre).

• Reworked the GitHub CI testing workflow to perform version checks as well as tests with pytest-cov (by @Zeitsperre).

• The xncml package has been significantly restructured to improve maintainability (by @bzah and @Zeitsperre):

  • The package now complies with both PEP 517 and PEP 621 (pyproject.toml and newer metadata definitions).

  • The build system has been migrated from setuptools to flit.

  • CHANGES.rst has been replaced with CHANGELOG.rst to follow the keepachangelog format.

  • Code linting and formatting now exclusively uses ruff (in lieu of black, isort, and others).

  • The package has adopted a src layout.

  • Documentation now uses sphinx-apidoc to generate API documentation on build and the module layout is now navigable on the documentation page.

  • The xncml package now uses pytest for testing, and the test suite has been migrated to pytest from unittest.

  • The project now uses Trusted Publisher to sign and verify releases.

0.4.0 (2024-01-08)

• Add support for <EnumTypeDef> (by @bzah).

• Update XSD schema and dataclasses to latest version from netcdf-java to add support for unsigned types (by @bzah).

• Add support for scalar variables (by @bzah).

• [fix] empty attributes are now parsed into an empty string instead of crashing the parser (by @bzah).

0.3.1 (2023-11-10)

• Add support for Python 3.12

• Drop support for Python 3.8

0.3 (2023-08-28)

• Add add_aggregation and add_variable_agg to Dataset class (by @huard).

• Add add_scan to Dataset class (by @huard).

• Closing the dataset returned by open_ncml will close the underlying opened files (by @huard).

• Add Dataset.from_text classmethod to create a Dataset from an XML string (by @huard).

0.2 (2023-02-23)

• Implement Dataset.rename_dataset_attribute (by @huard).

• Allow empty Dataset creation (by @huard).

• Add support in Dataset for NcML documents using the ncml namespace (by @huard).

• Implement Dataset.to_cf_dict method to export CF-JSON dictionary (by @huard).

0.1 Initial release (2022-11-24)

• Manipulate NcML file: add & remove attributes, variables and dimensions. (by @andersy005).

• Implement open_ncml, which returns an xarray.Dataset built from an NcML. Note that Only a subset of the NcML syntax is supported. (by @huard).