Skip to main content
Version: v1.2.0


Dataflow operators process a sequence of input values to create an output sequence and appear as the components of a dataflow pipeline.

  • assert - evaluate an assertion
  • combine - combine parallel paths into a single output
  • cut - extract subsets of record fields into new records
  • drop - drop fields from record values
  • from - source data from pools, files, or URIs
  • fork - copy values to parallel paths
  • fuse - coerce all input values into a merged type
  • head - copy leading values of input sequence
  • join - combine data from two inputs using a join predicate
  • over - traverse nested values as a lateral query
  • put - add or modify fields of records
  • rename - change the name of record fields
  • sample - select one value of each shape
  • search - select values based on a search expression
  • sort - sort values
  • summarize - perform aggregations
  • switch - route values based on cases
  • tail - copy trailing values of input sequence
  • uniq - deduplicate adjacent values
  • where - select values based on a Boolean expression
  • yield - emit values from expressions