v1.14.0

• Added support for positional values as input for wrangles and columns + not_columns for read/write.

  wrangles:
    - wrangle:
         input: 0  # 0-indexed
         # or
         input:
           - 0 
           - 1
         # or
         input: '1:7' # support python slicing syntax

• Standardize: Fixed an error with the schema.

• Extract.AI: Added support for strict mode (true/false, default false). This enforces compliance to the schema provided, but may have a more limited range of schema support.

• Transpose: Added header_column parameter to choose the column that will become the new column headings by either name or position (0-indexed).

  • Possible minor breaking change: the default header_column is now the first column. To restore the old behaviour and not use a column, use header_column: null.

• Format.suffix/Format.prefix: Ensure values are appended as strings so other types such as numbers do not fail.

• Train.lookup: Enforce validation for duplicate keys.

• Matrix: Fixed a bug where it would fail when using the set() syntax with a column that contains non-hashable types.

• Lookup: Fixed a bug where it would fail for empty dataframes.

• file connector: Allow reading files as an object with base64 encoded data. To be used with the new file variable type for WranglesXL.

• Automatically retry wrangles on unexpected server errors.

• Raise a more specific error if backend queries respond with an invalid response body.

• Disable tests for macOS on ARM using python 3.10. Due to a downstream dependency this is currently not supported. Suggest upgrade to 3.11 or newer.

• Added a df variable to if conditions that contains the dataframe object. This can be used to set more sophisticated conditions based on the dataframe contents.