Skip to content

Change log

Version 1.1.0 (2023-12-12)

  • Fixes to tests

Version 1.0.9 (2023-12-12)

  • Added in new usdate data type which supports automatic date type coersion from US date formats (MM-DD-YYYY).
  • New COLLATE action permitting scripts which collate a column-wise list of fields into an ordered array data type.
  • Updated all dependencies, which necessitated a bump to the minimum supported Python version being 3.9.
  • New tutorial to demonstrate the COLLATE and usdate features.
  • Ray settings for initialisation will now be more forgiving. You don't need to provide default values.

Version 1.0.8 (2023-08-10)

  • CategoryModel terms now use StrictBool to avoid Pydantic's liberal interpretation of booleans.

Version 1.0.7 (2023-08-09)

  • Minor fix.

Version 1.0.6 (2023-08-09)

  • Minor fix.

Version 1.0.5 (2023-08-09)

  • Additional ambiguity checks for category term edge case where source or destination fields can share category names.

Version 1.0.4 (2023-08-08)

  • Ambiguity checks for string blank space. If source data includes, should not be removed to preserve original structure.

Version 1.0.3 (2023-08-07)

  • Disambiguation where schema subject and object field categories have the same name.

Version 1.0.2 (2023-07-05)

  • Dependency updates.

Version 1.0.1 (2023-07-05)

  • Permitting nrow limit on Parquet files.

Version 1.0.0 (2023-05-10)

This version shares some features with the previous version, but is a complete refactoring and conceptual redesign. It is not backwardly compatible. Future versions will maintain compatability with this one.

  • Separated data models from schema models so that crosswalks are schema-to-schema.
  • Complete revision of the API into four discrete Definition classes, SchemaDefinition, DataSourceDefinition, CrosswalkDefinition and TransformDefinition.
  • Removed filters and actions that are no longer relevant (including REBASE, and MERGE).
  • Simplified CATEGORISE since it no longer requires deriving terms as part of the crosswalk.
  • Crosswalks are designed to support continuous integration.
  • Pydantic models are more transparent via each Definition's .get property.
  • Refactored Pandas to support Modin and Ray for data >1 million rows.
  • Mime type support for data sources in Parquet and Feather.
  • Rewrote documentation in MKDocs from Sphinx.
  • Revised all tutorials and documentation.

Legacy version history

  • 0.6.2: Fix for parsing ambiguity errors, plus Excel row-count exceeded on save.
  • 0.6.1: Minor correction for row count.
  • 0.6.0: Ensuring consistent optional column 'name' change for header row, plus new row-count in input and working data.
  • 0.5.0: Breaking API changes, including a complete rebuild to support Pydantic and type annotations.
  • 0.3.1: Minor documentation corrections & updated Manifest.
  • 0.3.0: Introduced Morph table restructuring functions.
  • 0.2.0: Refactored Action wrangling functions to support new drop-in actions.
  • 0.1.1: Minor bug fixes.
  • 0.1.0: Initial release.