CatIndices

A Julia package for concatenating, growing, and shrinking arrays in ways that allow control over the resulting axes.

Usage

BidirectionalVector

These vectors can grow or shrink from either end, and the axes update correspondingly. In this demo, pay careful attention to the axes at each step:

julia> using CatIndices

julia> v = BidirectionalVector(rand(3))
CatIndices.BidirectionalVector{Float64} with indices CatIndices.URange(1,3):
 0.32572
 0.250426
 0.834728

julia> append!(v, rand(2))
CatIndices.BidirectionalVector{Float64} with indices CatIndices.URange(1,5):
 0.32572
 0.250426
 0.834728
 0.388788
 0.282573

julia> prepend!(v, rand(3))
CatIndices.BidirectionalVector{Float64} with indices CatIndices.URange(-2,5):
 0.992902
 0.849368
 0.189849
 0.32572
 0.250426
 0.834728
 0.388788
 0.282573

julia> pop!(v)
0.28257294456774673

julia> axes(v)
(CatIndices.URange(-2,4),)

julia> popfirst!(v)
0.9929020233076613

julia> axes(v)
(CatIndices.URange(-1,4),)

deleteat! and insert! are not supported, since it is unclear whether it should shrink/grow from the beginning or end. To eliminate many items at the beginning or end of the vector, this package exports deletehead!(v, n) and deletetail!(v, n).

Concatenation

This is still mostly a TODO. For one-dimensional arrays (AbstractVectors), PinIndices provides a convenient interface for specifying which indices "win":

julia> v = vcat(1:3, PinIndices(4:5), 6:10)
10-element OffsetArray(::Array{Int64,1}, -2:7) with eltype Int64 with indices -2:7:
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10

julia> v[1]
4

The array wrapped in PinIndices keeps its own indexes, and everything else adjusts to compensate.