CHANGELOG.md

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

6.1.3 - 2026-03-07

Changed

• Switch to uv (including build backend).

6.1.2 - 2026-03-03

Fixed

• Fix rics kernel sometimes producing invalid kernel names.

6.1.1 - 2026-01-17

Fixed

• Various docstrings.
• Expose the MakeDict type (was already "exposed" through rics.collections.dict.MakeType).

6.1.0 - 2025-12-08

Added

• New class strings.ReprFormatter.
• Function strings.format_kwargs(), backed by strings.ReprFormatter.

Changed

• The misc.format_kwargs() now delegates to strings.format_kwargs(), and will be deprecated in 0.7.0.

6.0.4 - 2025-10-25

Added

• Python 3.14 is now fully tested and supported in CI/CD.

Fixed

• Fix handling format_kwargs(max_value_length=0) and clarify docstring.

6.0.3 - 2025-07-30

Fixed

• In get_public_module():
  • Handle objects defined in '__main__'.
  • Handle objects with no public exports (fall back to full private path).

6.0.2 - 2025-07-27

Fixed

• CLI fixes for rics kernel:
  • Ensure that VIRTUAL_ENV is unset (not just blank) when running commands; caused Poetry to purge venv.
  • Increase package install timeout.

6.0.1 - 2025-06-25

Fixed

• Add missing space in format_seconds(t <= 60.0) output, e.g. '12ms' → '12 ms'.

6.0.0 - 2025-06-20

Changed

• Make pandas an optional dependency.
• Args run_multivariate_test(time_per_candidate, plot) are now kw-only.

Deprecated

• The rics.envinterp module.
• The mics.interpolate_environment_variables() function.

5.1.2 - 2025-06-14

Fixed

• Fix crashes when using python -OO.

5.1.1 - 2025-06-12

Fixed

• CLI updates
  • Fix immediate crash when seaborn is missing.
  • Added rics timeit --no-plot (default False) to allow timeit without seaborn.
• CLI rics kernel updates
  • Fix uv detection.
  • Fix nested venv issues.

5.1.0 - 2025-05-29

Added

• Add logs.FORMAT_SEC, rename FORMAT -> FORMAT_MS.
• New class logs.LoggingSetupHelper.
• New functions logs.get_logger() and convert_log_level().
• New naive functions strings.camel_to_snake() and snake_to_camel().
• New class types.LiteralHelper[T].
• New wrapper functions types.verify_literal() and verify_enum().
• New module rics.env:
  • Added read-functions for primitive types; read_bool(), read_int, read_enum().
  • Added types.LiteralHelper[T].read_env().
• New function strings.str_as_bool().
• New module rics.click.
• The MultiCaseTimer now supports callable test_data args to generate data.
• Finished new rics.performance plotting implementation (support facets).

Changed

• Update basic_config.basic_config(): Allow and handle level=None to avoid logging from root.

• The misc.get_by_full_name() function now supports reading member attributes. Uses entrypoint syntax, e.g. pandas:DataFrame.sum.

• Moved some functions to new rics.env module:

  • Moves implementation rics.envinterp -> rics.env.interpolation.
  • Moved implementation of misc.interpolate_environment_variables() -> env.interpolation.replace_in_string().

  Aliases above will be deprecated in 0.6.0 and removed in 0.7.0.

• Added strings.format_seconds(full=False); round to nearest minute for durations above four hours.

Fixed

• Calling MultiCaseTimer.run(number=<int>) no longer crashes.
• Calling tname() with prefix_classname=True when arg is an inherited instance method (e.g. pd.Series().describe) now uses the child name. This parent is still shown for methods of uninitialized child types (e.g. pd.Series.describe).

5.0.1 - 2025-02-01

Fixed

• Fix RTD build. Docs for 5.0.0 are not available, but 5.0.1 docs are identical.

5.0.0 - 2025-02-01

Added

• Argument include_module=False for function misc.tname().
• New function plotting.percentage_ticks()

Fixed

• Handling of gh-101860 (new @property.__name__) under 3.13 in misc.tname() and misc.get_public_module().

Removed

• Module rics.ml.time_split. use the time-split package instead.

4.1.1 - 2024-06-29

Fixed

• Fix handling of paths.parse_any_path() bool-type postprocessors (raise on False).

4.1.0 - 2024-05-25

Added

• New module rics.paths; paths.parse_any_path(), derived functions any_path_to_str() and any_path_to_path().
• New module rics.strings. Moved format_perf_counter() and format_seconds().
• New function strings.format_bytes().
• New function rics.collections.dicts.format_changed_keys().

Deprecated

• Module rics.ml.time_split. use the time-split package instead.
• Functions rics.performance.format_perf_counter() and format_seconds(). Use rics.strings-functions instead.

4.0.1 - 2024-03-21

Fixed

• Fix RTD build. Docs for 4.0.0 are not available, but 4.0.1 docs are identical.

4.0.0 - 2024-03-21

Added

• Python 3.12 is now fully tested and supported in CI/CD.
• New support module time_split.integration.split_data.
• New time_split integration: integration.polars.split_polars.

Changed

• Python minimum version is now 3.11 (was 3.8).
• The time_split Pandas integration no longer supports the inclusive=left|right|neither argument (now always left).
• Updates to rics.performance:
  • Functions and classes are now generically typed.
  • Implement optional progress bars.
  • Added new plotting backend (experimental).
  • Select suitable time unit automatically when plotting (was: milliseconds).

Removed

• Module rics.mapping was previously deprecated and has now been removed.
• Module rics.pandas was previously deprecated and has now been removed.
• Module rics.strings (use str methods instead).

3.3.0 - 2024-02-11

Added

• The get_by_full_name() function has two new optional arguments instance_of and subclass_of, which may be used to ensure correct return types.
• The unflatten_dict() function now supports tuple keys.
• The logs.disable_temporarily() function now support passing loggers by name.

Fixed

• Print parent of inner class in misc.tname() when prefix_classname=True.
• Prefer later folds when using frequency-based schedule and after arguments (ml.time_split, #313).

3.2.0 - 2023-10-17

Added

• Add new step argument to ml.time_split splitting functions.
• Add new plot.REMOVED_FOLD_STYLE attribute to time_split.settings.

Fixed

• Update, fix and clarify some documentation issues.
• The misc.tname()-function now handles functools.partial properly (new argument attrs='func').

3.1.0 - 2023-10-13

Added

• New module rics.ml.time_split; successor to pandas.TimeFold.
• Function collections.dicts.unflatten_dict(); inverse of flatten_dict().
• Function misc.get_public_module(); resolve public module name of an object, optionally looking for reexports.
• Function misc.format_kwargs(); pretty-print keyword arguments.

Changed

• The flatten_dict()-function now takes a string_fn-argument (default=str).
• Update format used for durations in rics.performance.

Deprecated

• The rics.pandas.TimeFold class. Replaced by rics.ml.time_split.

Fixed

• Handle LoggerAdapter in logs.disable_temporarily().
• Handle improve handling of '{}' (e.g. Python fstrings) in envinterp.
• Handle properties in misc.tname.

3.0.1 - 2023-03-25

Changed

• Deprecated rics.mapping.

Fixed

• Fixed some minor TimeFold issues.
• TimeFold.make_sklearn_splitter.split and TimeFold.plot now handle datetime-like iterables correctly.

3.0.0 - 2023-03-09

Added

• The score_functions.disabled() function, used to force a Mapper to work in override-only mode.
• An optional argument for_value to the heuristic_functions.value_fstring_alias() function.
• The Mapper.copy()-method now accepts keyword override arguments.
• The envinterp module and misc.interpolate_environment_variables() function (replaces read_env_or_literal())

Changed

• Add n_splits argument to TimeFold (scikit-learn compatibility).
• Raise AmbiguousScoreError if match scores do not allow deterministic mapping.
• Add a small positional penalty in HeuristicScore, to favor alias functions that are defined early.
• Add a small positional penalty in modified_hamming, to favor candidates that are defined early.
• Reduce default value of Mapper.min_score (from 1.0 to 0.9).
• Updated Mapper verbosity flag to be consistent. Now called verbose_logging everywhere.

Fixed

• The heuristic_functions.candidate_fstring_alias() function may now properly use value and context placeholders.
• Properly raise exceptions in [value/candidate]_fstring_alias for invalid fstring arguments.
• Added missing checks in Mapper.__eq__().

Removed

• The misc.read_env_or_literal() function has been replaced by rics.misc.interpolate_environment_variables().

2.1.0 - 2023-02-11

Changed

• Update log levels in logs.basic_config().
• Move legend outside figure in rics.performance.plot_run()
• Only show the best overall candidates (per dataset) when using mtimeit / run_multivariate_test() by default.
• Add optional ax argument to TimeFold.plot().

Fixed

• Fixed behaviour of TimeFold.plot() in Jupyter notebooks. This function now returns an Axes object instead of a Figure, which prevents figures from being shown twice in notebooks. Technically a breaking changes, but will be treated as a bugfix.
• Adjust figure height based on number of folds in TimeFold.plot(), unless figsize is given.
• Suppress fold log messages in TimeFold.plot().

2.0.0 - 2022-11-30

Added

• The Mapper.copy method.

Changed

• Flatten/remove rics.utility; promote all modules to top-level members.
• Mapper.apply now require context with InheritedKeysDict-type overrides.

Fixed

• Fixed exact value-candidate match short-circuiting in HeuristicScore.

Removed

• All rics.translation modules. Use id-translation instead.

1.0.1 - 2022-11-26

Changed

• Deprecate rics.translation. Use id-translation instead.

1.0.0 - 2022-11-26

Added

• The logs.disable_temporarily function (returns a context manager).
• Permit creation of derived classes with Translator.from_config.
• Flesh out description of the Translatable type variable.
• Ability for the perftest CLI program to create and run dummy tests.
• Utility method utility.pandas.TimeFold.iter: Create temporal k-folds from a heterogeneous DataFrame.
• Make SqlFetcher engine creation more customizable.
  • Add SqlFetcher.create_engine, which my be overridden.
  • Add optional engine_kwargs init argument.

Changed

• Change definition of Mapper.context_sensitive_overrides; now also True for blank overrides regardless of type.
• Discard useless (no-source) fetchers in MultiFetcher.
• The SqlFetcher class now always includes the engine information in log messages.
• The utility.misc.tname-method now also considers __name__ before falling back to __class__.__name__.
• The utility.configure_stuff-method now also sets the pandas float_format option.
• Renamed perftest CLI program to mtimeit (was: rics-perf).
• Feature gate all CLI functionality behind the new cli extra.

Fixed

• Properly handle empty whitelists in SqlFetcher. Used to treat empty whitelist as no whitelist.

0.17.0 - 2022-09-29

Added

• Option to print classname for functions when using misc.tname.
• Property Translator.sources
• The Translator.map_scores method. May be used to fetch the raw name-to-source match score matrix.

Changed

• Print less duplicate information in Mapper.compute_scores and Mapper.apply.
• Make Mapper.compute_scores and Mapper.apply respect the iteration order of its arguments when there are match conflicts.
• Improve messaging when translation is aborted due to mapping failure.

0.16.1 - 2022-09-23

Changed

• Switch from toml to tomli for reading Translator config files.
• Improve documentation for troubleshooting mapping issues

Fixed

• The utility.logs.basic_config-function now properly uses the format and datefmt arguments.

0.16.0 - 2022-09-12

Changed

• Improved handling of NaN IDs. Move all NaN handling in AbstractFetcher to the Translator.
• Raise a MappingError for ambiguous name-to-source mapping.

Fixed

• The AbstractFetcher no longer treats zero-length ID collections as a fetch-all instruction.
• Cast float to int when extracting IDs from a pandas DataFrame or Series.

0.15.3 - 2022-09-05

Fixed

• Respect index order in MatchScores.to_directional_mapping.

0.15.2 - 2022-09-05

Fixed

• Make sure Mapper.compute_scores score respects given value/candidate order.

0.15.1 - 2022-09-02

Changed

• Added duplicate_key_action arg to utility.dicts.reverse_dict

Fixed

• Duplicate derived names for candidates in for perftests.
• Improve handling for optional Format blocks (#52). Nested optional blocks are not supported.

0.15.0 - 2022-08-01

Added

• Verbosity-flags to control high-volume mapping function invocation logging.
• Mapping primer page to the docs.
• Multiple improvements to the Mapper class through a new mapping.support module:
  • Improved logging capabilities.
  • Force respecting cardinality argument in Mapper.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_term argument to score_functions.modified_hamming.
• The utiltiy.collections.as_list function for wrapping in/casting to list
• Support for translating an attribute of translatable in Translator.translate.
  • Names may be inherited from the parent (ie a pandas Index may inherit the name of the series)
• Cookbook recipes for translating dict keys and Pandas index types.
• Translation of pandas.Index types.
• The translation.testing module.
• Experimental and hacky implementation of translation for nested sequences.
• Entry point rics-perf for multivariate performance testing, taking candidates from ./candidates.py and test case data from ./test_data.py.

Changed

• Rename Translator.map_to_sources -> map.
• Simplify Translator.translate signature.
  • Names must now be explicit names or None (use heuristics to filter names).
  • Simplify multiple docstrings.
• Permit Translator instances to be created with explicit fetch data. Translations will be generated based on the inputs by using a TestFetcher instance. 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 Mapper
• The fetching.support.from_records method. Fixes spurious exceptions from PandasFetcher (#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_candidate now used correctly.
• Filter out NaN values in AbstractFetcher.

0.14.0 - 2022-07-17

Changed

• Added home page shortcuts.
• Rename 'default_translations' and 'default' arguments to 'default_fmt_placeholders' .

Fixed

• Remove placeholder limitation on default translation format.
• Fixed issue when copying a Translator with translation and/or Format overrides.
• A number of docstring and under-the-hood fixes.

0.13.0 - 2022-07-10

Bump development status to Development Status :: 3 - Alpha on PyPi. Switched to the PyData Sphinx theme and enabled automatic summaries.

Changed

• Name of OfflineError changed to ConnectionStatusError.
• Moved Cardinality to the mapping namespace.
• Move utility.perf up one level.
• Swapped generic TypeVar order for IdType and SourceType to match the name -> source -> ID hierarchy.

Added

• Implement override functions in Mapper.apply.
  • Also: Partial implementation of override functions for name-to-source mapping in Translator.translate.
• Implement reverse translations. Added reverse argument to Translator.translate to translate from translations back to IDs.
• An option maximal_untranslated_fraction to raise an error if translation fails for too many IDs in Translator.translate.
• Make it possible to initialize Fetchers from arbitrary packages in Translator.from_config.
• Make it possible configure ScoreFunctions, FilterFunctions and AliasFunctions from arbitrary modules . (still defaults to package functions).
• The py.typed marker (PEP-561 compliance).
• Additional types-modules for typehint imports.

Fixed

• Numerous doc fixes.

0.12.2 - 2022-07-04

Fixed

• Fixed chained alias functionality for HeuristicScore.

0.12.1 - 2022-07-04

Botched release.

0.12.0 - 2022-07-03

Changed

• Changed like_database_table from score function to alias function.
• Thread safety: Change 'candidates' from init/property to Mapper.apply arg.
• Remove store/restore serialize argument. Default path is now None for Translator.store.

0.11.1 - 2022-07-02

Documentation and docstring fixes.

Added

• A TOML Configuration section to docs.

0.11.0 - 2022-07-01

Added

• Convenience functions for (de)serializing Translator instances.

Fixed

• Broken docs.

0.10.2 - 2022-07-01

Try to restore the translation API docs.

0.10.1 - 2022-06-30

Fixed

• The Translator retrieving incorrect sources in offline mode.

0.10.0 - 2022-06-30

Refactor scoring functions.

Added

• The HeuristicScore class, which enables filter-based short-circuiting and alias heuristics for score functions.

Removed

• The filter_functions.score_with_heuristics function; replaced by the more general HeuristicScore class.

Changed

• Make translation.factory module public.
• Rename MappingScoreFunction -> ScoreFunction.
• Add a third mandatory context parameter to all mapping functions. The source argument has been removed.

Fixed

• Fix handling logging of duplicate source discovery in MultiFetcher.
• Allow empty fetching section in main config file when extra_fetchers are given.

0.9.0 - 2022-06-28

Added

• Short-circuiting logic to filter_functions.score_with_heuristics.

Fixed

• Some scoring stuff and related logging.

0.8.1 - 2022-06-28

Fixed

• Fix unwanted case-sensitivity in filter_functions.score_with_heuristics.
• Allow user-defined cardinality in config files.
• Missing Cardinality.parse(str) case.

0.8.0 - 2022-06-28

Added

• Support for fetching using multiple fetchers using the MultiFetcher class (#44).
• The fetching.types module.
• The Fetcher interface. The old Fetcher class in now called AbstractFetcher.
• The action_level.ActionLevel enum type.

Changed

• Refactor score and filter functions (mapping-module) to increase flexibility and cover more use cases (#57, #58).

0.7.0 - 2022-06-21

Added

• The InheritedKeysDict class.
• Integration of Mapper in the Fetcher for programmatic placeholder-mapping.
• Filtering logic for Mapper.
• The plottting.pi_ticks-function to plot an (X-)axis in terms of PI.
• Dvdrental testcase and docs running against a docker DB (not in CI)

Changed

• More plotting functions.
• Improved dependency management. Added new dependency groups translation and plotting to be installed as extras.

0.6.0 - 2022-06-17

Added

• Copy method for Translator.copy(**overrides).

Changed

• Switch from YAML to TOML for Translator.config.

0.5.0 - 2022-06-14

A large number of changes, bugfixes and stability improvements. Only the most important ones are listed here.

Added

• Implement shared default translations (#31).
• Implement in-place translation for sequences (list, pandas.Series, numpy.array).
• Allow alternative use of translation format with an ID placeholder, even if the main doesn't include it.
• Add an option to run Translator.store with explicit data to cache.
• Implement alternative format for unknown IDs.
• Dict utility methods in rics.utility.collections.

Changed

• SqlFetcher: Break out potentially expensive operations into overridable methods
• Replace suffix _log_level -> _level in basic_config

Fixed

• Fix name extraction for pandas.Series

0.4.0 - 2022-03-24

Added

• The rics.utility.plotting module.

0.3.2 - 2022-03-17

Added

• The Translator.fetch function for retrieving IDs.

Fixed

• Revert bad commit.
• Fix an issue in PandasFetcher.
• Some SqlFetcher warning messages.

0.3.1 - 2022-03-17

Fixed

• Readme links.

0.3.0 - 2022-03-17

Changed

• Update get_local_or_remote - remote root mandatory, default to cwd for local root.

Added

• Make it possible to use predicates for names and ignore_name in Translator.translate.
• Add Translator.map_to_sources to get name-to-source mapping without translating data.

Fixed

• Translation using formats which don't include the id placeholder.

0.2.0 - 2022-03-15

Changed

• Update Fetcher return format from dict-of-lists to matrix (performance optimization).

Added

• Database table discovery (gathers metadata like size and columns) for SqlFetcher.
• Add selection logic (used to always fetch everything) for SqlFetcher.
• Implement translations of pandas.Series.
• The rics.utility.perf package for multivariate (data x functions) performance testing.

Fixed

• Fix fetching only required placeholders in SqlFetcher.

0.1.0 - 2022-03-12

First release on PyPI!

Added

• The rics.utility package for various common operations.
• The rics.translation package suite for translating IDs into human-readable labels.
• The rics.mapping package for linking elements in multiple directions.
• The rics.cardinality package; enum types for 1:1, 1:N, N:1, and M:N.