FlexiJoins.jl is a fresh take on joining tabular or non-tabular datasets in Julia.

From simple joins by key, to asof joins, to merging catalogs of terrestrial or celestial coordinates – FlexiJoins supports any usecase.

Defining features of FlexiJoins that make it flexible:

• Wide range of join conditions:
  • by key, so-called equi-join
  • by distance
  • by a comparison predicate: one of <, <=, ==, >=, >
  • all matches or only the closest match
  • by an interval predicate: one of ∈, ⊆, ⊊, ⊋, ⊇, !isdisjoint
  • combinations of the above
• All kinds of joins, as in inner/left/right/outer
• Results can either be a flat list, or grouped by the left/right side
• Lots of dataset types transparently supported: various arrays, dictionaries, tables
• And more! See examples.

With all these features, FlexiJoins is designed to be easy-to-use and fast:

• Uniform interface to all functionaly
• Performance close to other, less general, solutions: see benchmarks comparing with SplitApplyCombine.jl and DataFrames.jl
• Extensible in terms of both new join conditions and more specialized algorithms

Examples that showcase main features:

innerjoin((objects, measurements), by_key(:name))

leftjoin((O=objects, M=measurements), by_key(:name); groupby=:O)

innerjoin((M1=measurements, M2=measurements), by_key(:name) & by_distance(:time, Euclidean(), <=(3)))

innerjoin(
	(O=objects, M=measurements),
	by_key(:name) & by_pred(:ref_time, <, :time);
	multi=(M=closest,)
)

Documentation with explanations and more examples is available as a Pluto notebook. Please report bugs/issues on github, but direct usage questions to the discourse topic.