0.15.0 (August 01, 2022)#
Added#
Verbosity-flags to control high-volume mapping function invocation logging.
Mapping primer page to the docs.
Multiple improvements to the
Mapperclass through a newmapping.supportmodule:Improved logging capabilities.
Force respecting
cardinalityargument inMapper.apply(fewer errors are raised).Allow users to retrieve raw score matrices (
Mapper.compute_scores).Improved documentation and added new links is multiple places.
Possibility to override select filtering logic;
SqlFetcher.selection_filter_type.An optional
add_length_ratio_termargument toscore_functions.modified_hamming.The
utiltiy.collections.as_listfunction for wrapping in/casting to listSupport for translating an attribute of
translatableinTranslator.translate.Names may be inherited from the parent (ie a pandas
Indexmay inherit the name of the series)
Cookbook recipes for translating dict keys and Pandas index types.
Translation of
pandas.Indextypes.The
translation.testingmodule.Experimental and hacky implementation of translation for nested sequences.
Entry point
rics-perffor multivariate performance testing, taking candidates from./candidates.pyand test case data from./test_data.py.
Changed#
Rename
Translator.map_to_sources->map.Simplify
Translator.translatesignature.Names must now be explicit names or None (use heuristics to filter names).
Simplify multiple docstrings.
Permit
Translatorinstances to be created with explicit fetch data. Translations will be generated based on the inputs by using aTestFetcherinstance. Functionality in this mode is limited.Performance testing figures updated; now shows best result as well.
Removed#
An unnecessary restriction on runtime override functions in
MapperThe
fetching.support.from_recordsmethod. Fixes spurious exceptions fromPandasFetcher(#99).Dunder
Mapper.__call__.Expected runtime checks for perftests.
Fixed#
An issue when using integers as explicit names to translate.
It is now possible to use one name per element when translating sequences.
Perftest argument
time_per_candidatenow used correctly.Filter out
NaNvalues inAbstractFetcher.