convert Arrow-flavored eltypes to Julia-flavored eltypes on copy

[jrevels]

This improves consistency with the copy path for nullable columns, and is also important for avoiding unexpected situations with e.g. DataFrames operations that call copy internally.

Before:

julia> using Arrow, Dates, UUIDs, DataFrames

julia> tt = begin
           t = (a = [Nanosecond(0), Nanosecond(1)], b = [uuid4(), uuid4()], c = [missing, Nanosecond(1)])
           io = IOBuffer()
           Arrow.write(io, t)
           seekstart(io)
           Arrow.Table(io)
       end
Arrow.Table: (a = [Nanosecond(0), Nanosecond(1)], b = UUID[UUID("3258e377-fd8c-4305-9e6a-e1348c3ef362"), UUID("05473e64-0ac3-4951-a668-43d195813429")], c = Union{Missing, Nanosecond}[missing, Nanosecond(1)])

julia> copy(tt.a)
2-element Array{Arrow.Duration{Arrow.Flatbuf.TimeUnitModule.NANOSECOND},1}:
 Arrow.Duration{Arrow.Flatbuf.TimeUnitModule.NANOSECOND}(0)
 Arrow.Duration{Arrow.Flatbuf.TimeUnitModule.NANOSECOND}(1)

julia> copy(tt.b)
2-element Array{UInt128,1}:
 0x3258e377fd8c43059e6ae1348c3ef362
 0x05473e640ac34951a66843d195813429

julia> copy(tt.c)
2-element Array{Union{Missing, Nanosecond},1}:
 missing
 1 nanosecond
 
 julia> transform(identity, DataFrame(tt))
2×3 DataFrame
 Row │ a                        b                                  c
     │ Duration…                UInt128                            Nanoseco…?
─────┼──────────────────────────────────────────────────────────────────────────
   1 │ Duration{NANOSECOND}(0)  94652081088113183412278002734480…  missing
   2 │ Duration{NANOSECOND}(1)  16916473316481761014725308728310…  1 nanosecond

after:

julia> using Arrow, Dates, UUIDs, DataFrames

julia> tt = begin
           t = (a = [Nanosecond(0), Nanosecond(1)], b = [uuid4(), uuid4()], c = [missing, Nanosecond(1)])
           io = IOBuffer()
           Arrow.write(io, t)
           seekstart(io)
           Arrow.Table(io)
       end
Arrow.Table: (a = [Nanosecond(0), Nanosecond(1)], b = UUID[UUID("43650671-ccf7-46cb-8fa9-a0c63b7f2e7b"), UUID("828d37d7-975f-4d6e-89bb-3a2f48ed1b10")], c = Union{Missing, Nanosecond}[missing, Nanosecond(1)])

julia> copy(tt.a)
2-element Array{Nanosecond,1}:
 0 nanoseconds
 1 nanosecond

julia> copy(tt.b)
2-element Array{UUID,1}:
 UUID("43650671-ccf7-46cb-8fa9-a0c63b7f2e7b")
 UUID("828d37d7-975f-4d6e-89bb-3a2f48ed1b10")

julia> copy(tt.c)
2-element Array{Union{Missing, Nanosecond},1}:
 missing
 1 nanosecond
 
 julia> transform(identity, DataFrame(tt))
2×3 DataFrame
 Row │ a              b                                  c
     │ Nanoseco…      UUID                               Nanoseco…?
─────┼────────────────────────────────────────────────────────────────
   1 │ 0 nanoseconds  43650671-ccf7-46cb-8fa9-a0c63b7f…  missing
   2 │ 1 nanosecond   828d37d7-975f-4d6e-89bb-3a2f48ed…  1 nanosecond

[codecov]

Codecov Report

Merging #98 (71ee79c) into main (f92592c) will increase coverage by 0.29%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main      #98      +/-   ##
==========================================
+ Coverage   83.68%   83.98%   +0.29%     
==========================================
  Files          23       23              
  Lines        2673     2673              
==========================================
+ Hits         2237     2245       +8     
+ Misses        436      428       -8     

Impacted Files	Coverage Δ
src/arraytypes/primitive.jl	80.00% <100.00%> (+6.00%)	⬆️
src/arraytypes/arraytypes.jl	90.00% <0.00%> (+1.11%)	⬆️
src/eltypes.jl	85.16% <0.00%> (+1.69%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f92592c...71ee79c. Read the comment docs.