Pipelines can declare transforms to transform processed records.
Transforms are applied to single attributes and have access to entire records, i.e., all other attributes. Exemplary transforms are a tokenizer, type casts, or hash algorithms.
DataCater ships a set of pre-defined transforms for the most common use cases in operational transforms. In addition, DataCater supports custom Python transforms that can be defined inline in the YAML spec of pipelines or can be edited in the UI with full support for previews.