Bounds error when sorting a column after `select`

[George9000]

Using Julia 1.9.1 and DataFrames 1.5.0, a bounds error is thrown when attempting to sort a dataframe by a column following a select and reassignment operation. This is not reproducible in another column in the same dataframe. Despite scrutinizing the contents of the column throwing the error, no clues to source are found.
[version details in expandable section below code]

Example data to reproduce error is here.

using DataFrames, CSV

const datap = "."

# Bounds error
let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame)
    df = select(df, [:sku, :units])
    sort(df, :sku)
end


#Success
let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame)
    df = select(df, [:sku, :units])
    sort(df, :units)
end


#Success
let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame)
    df = select(df, [:sku, :units])
    sort(df.sku)
end


#Success
let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame)
    sort(df, :sku)
end



# CSV.read error: InexactError: trunc(Int64, NaN)
let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame; skipto = 8000, limit = 5000)
end


##  investigate :sku column ##
function skucategory(sku)
    if ismissing(sku)
        return missing
    elseif occursin(r"^\d+$", sku)
        return "digits"
    elseif occursin(r"^[\dS]+$", sku)
        return "S_sku"
    elseif occursin(r"^[[:alnum:]]+$", sku)
        return "alphanumeric"
    elseif occursin(r"\/", sku)
        return "slash"
    else
        return "other"
    end
end

let
    df = CSV.read(joinpath(datap, "examp.csv"), DataFrame)
    transform!(df, :sku => ByRow(skucategory) => :skucat)
    gdf = groupby(df, :skucat)
    combine(gdf, nrow)
end   

version info, project.toml, error output

julia> versioninfo()
Julia Version 1.9.1
Commit 147bdf428c (2023-06-07 08:27 UTC)
Platform Info:
  OS: macOS (arm64-apple-darwin21.6.0)
  CPU: 10 × Apple M1 Max
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-14.0.6 (ORCJIT, apple-m1)
  Threads: 5 on 8 virtual cores
Environment:
  JULIA_NUM_THREADS = 4
  JULIA_EDITOR = Emacs
  JULIA_PKG_DEVDIR = /Users/george/Documents/julia/dev
  
Status `~/Desktop/exampl/Project.toml`
  [336ed68f] CSV v0.10.11
  [a93c6f00] DataFrames v1.5.0 `https://github.com/JuliaData/DataFrames.jl.git#main`

# Bounds error
ERROR: BoundsError: attempt to access 217465-element Vector{Int64} at index [0]
Stacktrace:
  [1] setindex!
    @ ./array.jl:969 [inlined]
  [2] _sort!(v::Vector{Int64}, a::Base.Sort.MissingOptimization{Base.Sort.BoolOptimization{Base.Sort.Small{10, Base.Sort.InsertionSortAlg, Base.Sort.IEEEFloatOptimization{Base.Sort.IsUIntMappable{Base.Sort.Small{40, Base.Sort.InsertionSortAlg, Base.Sort.CheckSorted{Base.Sort.ComputeExtrema{Base.Sort.ConsiderCountingSort{Base.Sort.CountingSort, Base.Sort.ConsiderRadixSort{Base.Sort.RadixSort, Base.Sort.Small{80, Base.Sort.InsertionSortAlg, Base.Sort.ScratchQuickSort{Missing, Missing, Base.Sort.InsertionSortAlg}}}}}}}, Base.Sort.StableCheckSorted{Base.Sort.ScratchQuickSort{Missing, Missing, Base.Sort.InsertionSortAlg}}}}}}}, o::Base.Order.Perm{Base.Order.ForwardOrdering, Vector{Union{Missing, String31}}}, kw::NamedTuple{(:lo, :hi, :allow_legacy_dispatch), Tuple{Int64, Int64, Bool}})
    @ Base.Sort ./sort.jl:583
  [3] sort!
    @ ./sort.jl:2123 [inlined]
  [4] sort!(v::Vector{Int64}, lo::Int64, hi::Int64, #unused#::SortingAlgorithms.TimSortAlg, o::Base.Order.Perm{Base.Order.ForwardOrdering, Vector{Union{Missing, String31}}})
    @ SortingAlgorithms ~/.julia/packages/SortingAlgorithms/n1AWW/src/SortingAlgorithms.jl:488
  [5] sort!
    @ ./sort.jl:2121 [inlined]
  [6] _sortperm(df::DataFrame, a::SortingAlgorithms.TimSortAlg, o::Base.Order.Perm{Base.Order.ForwardOrdering, Vector{Union{Missing, String31}}})
    @ DataFrames ~/.julia/packages/DataFrames/4GaAs/src/abstractdataframe/sort.jl:592
  [7] sortperm(df::DataFrame, cols::Symbol; alg::Nothing, lt::typeof(isless), by::typeof(identity), rev::Bool, order::Base.Order.ForwardOrdering)
    @ DataFrames ~/.julia/packages/DataFrames/4GaAs/src/abstractdataframe/sort.jl:589
  [8] sortperm
    @ ~/.julia/packages/DataFrames/4GaAs/src/abstractdataframe/sort.jl:577 [inlined]
  [9] #sort#378
    @ ~/.julia/packages/DataFrames/4GaAs/src/abstractdataframe/sort.jl:510 [inlined]
 [10] sort(df::DataFrame, cols::Symbol)
    @ DataFrames ~/.julia/packages/DataFrames/4GaAs/src/abstractdataframe/sort.jl:503
 [11] top-level scope
    @ REPL[5]:4


# CSV inexact error
ERROR: InexactError: trunc(Int64, NaN)
Stacktrace:
 [1] trunc
   @ ./float.jl:893 [inlined]
 [2] ceil(#unused#::Type{Int64}, x::Float64)
   @ Base ./float.jl:383
 [3] CSV.Context(source::CSV.Arg, header::CSV.Arg, normalizenames::CSV.Arg, datarow::CSV.Arg, skipto::CSV.Arg, footerskip::CSV.Arg, transpose::CSV.Arg, comment::CSV.Arg, ignoreemptyrows::CSV.Arg, ignoreemptylines::CSV.Arg, select::CSV.Arg, drop::CSV.Arg, limit::CSV.Arg, buffer_in_memory::CSV.Arg, threaded::CSV.Arg, ntasks::CSV.Arg, tasks::CSV.Arg, rows_to_check::CSV.Arg, lines_to_check::CSV.Arg, missingstrings::CSV.Arg, missingstring::CSV.Arg, delim::CSV.Arg, ignorerepeated::CSV.Arg, quoted::CSV.Arg, quotechar::CSV.Arg, openquotechar::CSV.Arg, closequotechar::CSV.Arg, escapechar::CSV.Arg, dateformat::CSV.Arg, dateformats::CSV.Arg, decimal::CSV.Arg, groupmark::CSV.Arg, truestrings::CSV.Arg, falsestrings::CSV.Arg, stripwhitespace::CSV.Arg, type::CSV.Arg, types::CSV.Arg, typemap::CSV.Arg, pool::CSV.Arg, downcast::CSV.Arg, lazystrings::CSV.Arg, stringtype::CSV.Arg, strict::CSV.Arg, silencewarnings::CSV.Arg, maxwarnings::CSV.Arg, debug::CSV.Arg, parsingdebug::CSV.Arg, validate::CSV.Arg, streaming::CSV.Arg)
   @ CSV ~/.julia/packages/CSV/OnldF/src/context.jl:647
 [4] #File#32
   @ ~/.julia/packages/CSV/OnldF/src/file.jl:222 [inlined]
 [5] read(source::String, sink::Type; copycols::Bool, kwargs::Base.Pairs{Symbol, Int64, Tuple{Symbol, Symbol}, NamedTuple{(:skipto, :limit), Tuple{Int64, Int64}}})
   @ CSV ~/.julia/packages/CSV/OnldF/src/CSV.jl:117
 [6] top-level scope
   @ REPL[6]:2

[bkamins]

Thank you for reporting.

This is unrelated to DataFrames.jl. The fundamental problem is:

o = Base.Order.Perm(Base.Order.ForwardOrdering(), df.sku)
a = DataFrames.SortingAlgorithms.TimSortAlg()
x = collect(1:nrow(df))
sort!(x, a, o) # this errors

This is triggered in SortingAlgorithms.jl / Base.

@LilithHafner - maybe there were some recent changes to these algorithms? The column sorted is Vector{Union{String31, Missing}}.

[LilithHafner]

@George9000, thank you for reporting, and @bkamins, thank you for @ ing me. I can confirm that this is my fault. I introduced a fast path for sorting unions with missing that is only safe when sorting an entire vector (a la sort!(v)) but TimSort uses an internal method of sort sort!(v::AbstractVector, lo::Integer, hi::Integer, alg::Alorithm, o::Ordering) that incorrectly dispatches to this fast path.

I will fix the restrict the dispatch to the fastpath in Base's sorting to unbreak this internal method. This fix should make it into 1.9.2 and 1.10.0.

I will also patch SortingAlgorithms so that TimSort calls a public method (which I did not break). That will come out soon.

[bkamins]

@LilithHafner - thank you for a prompt response. You are a top expert in this field and your expertise is highly appreciated here.

Could you please x-link this issue to the PR you would make (so that I can close this issue when things are sorted out 😄 ).

[LilithHafner]

I'm assuming you are referring to the field of "fixing bugs that Lilith introduced" 😄 Happy to help.

[bkamins]

I really appreciate your efforts to improve sorting in Julia!

[LilithHafner]

@George9000, would you please run ]up and try again? I think this is fixed in SortingAlgorithms v1.1.1

[bkamins]

Works for me. Is JuliaLang/julia#50171 needed to fix the issue (or it is just another enhancement, but DataFrames.jl is already fixed without it?)

[LilithHafner]

These are Base.Sort bugs so JuliaLang/julia#50171 is the real fix. JuliaCollections/SortingAlgorithms.jl#78 is a temporary shim so that SortingAlgorithms (and DataFrames by extension) is not broken before JuliaLang/julia#50171 is released.

[bkamins]

Ah - OK. Now I see it. Still - this issue can be closed then. Right?