Releases: bcdev/xrlint
Releases · bcdev/xrlint
0.4.1
Changes
- Added core rule
conventions
that checks for theConventions
attribute. - Added core rule
context-descr
that checks content description - Added core rule
var-descr
that checks data variable description - Renamed rules for consistency:
var-units-attrs
andvar-units
flags
intovar-flags
Fixes
- Fixed an issue that prevented recursively traversing folders referred
to by URLs (such ass3://<bucket>/<path>/
) rather than local directory
paths. (#39)
0.4.0
-
Fixed and enhanced core rule
time-coordinate
. `(#33) -
New xcube rule
no-chunked-coords
. (#29) -
New xcube multi-level dataset rules:
ml-dataset-meta
: verifies that a meta info file exists and is consistent;ml-dataset-xy
: verifies that the levels have expected spatial resolutions;ml-dataset-time
: verifies that the levels have expected time dimension, if any.
-
Now supporting xcube multi-level datasets
*.levels
:- Added xcube plugin processor
"xcube/multi-level-dataset"
that is used
inside the predefined xcube configurations "all" and "recommended".
- Added xcube plugin processor
-
Introduced method
Plugin.define_config
which defines a named plugin
configuration. It takes a name and a configuration object or list of
configuration objects. -
Changed the way how configuration is defined and exported from
Python configuration files:- Renamed function that exports configuration from
export_configs
intoexport_config
. - The returned value should be a list of values that can be
converted into configuration objects: mixedConfig
instances,
dictionary, or a name that refers to a named configuration of a plugin.
- Renamed function that exports configuration from
-
Other changes:
- Property
config
ofLinter
now returns aConfigList
instead
of aConfig
object. - Directories that are recognized by file patterns associated with a non-empty
configuration object are no longer recursively traversed. - Node path names now contain the dataset index if a file path
has been opened by a processor produced multiple
datasets to validate. - Changed type of
Plugin.configs
fromdict[str, Config]
to
dict[str, list[Config]]
. - Inbuilt plugin rules now import their
plugin
instance from
xrlint.plugins.<plugin>.plugin
module. JsonSerializable
now recognizesdataclass
instances and no longer
serializes property values that are also default values.- Pinned zarr dependency to be >=2.18, <3 until test
tests.plugins.xcube.processors.test_mldataset.MultiLevelDatasetProcessorTest
is adjusted orfsspec
's memory filesystem is updated. - Now making use of the
expected
property ofRuleTest
.
- Property
Full Changelog: v0.3.0...v0.4.0
0.3.0
- Added more rules
- Fixed problem where referring to values in modules via the form
"<module>:<attr>"
raised. (#21) - Introduced factory method
new_plugin
which simplifies creating plugin objects. - Refactored out new common mixin class
Operation
which reduces amount of code and simplifies testing of operation classesRule
,Processor
,Formatter
. - Improved overall test coverage.
- Switched to ruff as default linter and formatter.
Full Changelog: v0.2.0...v0.3.0
0.2.0
- Rule description is now your
RuleOp
's docstring
ifdescription
is not explicitly provided. - Supporting virtual plugins: plugins provided by Python
dictionaries with rules defined by theRuleOp
classes. - Added more configuration examples in the
examples
folder. - All
xcube
rules now have references into the
xcube dataset specification. - Introduced mixin classes
ValueConstructible
and
derivedMappingConstructible
which greatly simplify
flexible instantiation of XRLint's configuration objects
and their children from Python and JSON/YAML values. - Made all docstrings comply to google-style.
0.1.0
- Added CLI option
--print-config PATH
, see same option in ESLint - XRLint CLI now outputs single results immediately to console, instead only after all results have been collected.
- Refactored and renamed
CliEngine
intoXRLint
. Documented the class. new_linter()
now uses a config name arg instead of a bool arg.- Split example notebook into two
0.0.3
0.0.2
- more rules
- more tests
- config list item can be config name
- xcube is no longer default plugin
- no rules by default, but no rules configured is an error
- CLI exit code 1 with max warnings exceeded
- CLI exit code 1 with no files given
- new CLI options
--plugin
,--rule
,--init
- markdown rule reference in docs via
mkruleref.py
tool - using
files
config option to specify valid filename extensions, see here - using default filename extensions
["**/*.zarr", "**/*.nc"]
- using subset of minimatch syntax instead of the simple globs used by Python's
fnmatch
- allow the CLI's
FILES
args to contain directories, which are automatically recursively traversed - rule
coords-for-dims
works on dataset level - support for custom processors