bug fix

Author: sl-solution

src/abstractdataset/selection.jl

@@ -114,7 +114,7 @@ Select columns of `ds` based on `args...`. `args` can be any column selector: co
 
 It modifies the data set in-place. See [`select`](@ref) if a copy of selected columns is desired.
 """
-function select!(ds, @nospecialize(args...))
+function select!(ds::Dataset, @nospecialize(args...))
     selected_cols_all = normalize_select(index(ds), args...)
     selected_cols = setdiff!(selected_cols_all[1], selected_cols_all[2])
     unwanted_cols = setdiff(1:ncol(ds), selected_cols)

src/byrow/row_functions.jl

@@ -1142,8 +1142,8 @@ end
 function row_join(ds::AbstractDataset, cols = :; threads = true, delim::AbstractString = ",", last::AbstractString = "")
     colsidx = multiple_getindex(index(ds), cols)
 
-    max_line_size = maximum(byrow(ds, sum, colsidx, by = y->length(__STRING(y)), threads = threads))
-    max_line_size += length(delim)*(length(colsidx)) + length(last)+1
+    max_line_size = maximum(byrow(ds, sum, colsidx, by = y->ncodeunits(__STRING(y)), threads = threads))
+    max_line_size += ncodeunits(delim)*(length(colsidx)) + ncodeunits(last)+1
     init0 = Matrix{UInt8}(undef, max_line_size, nrow(ds))
     curr_pos = ones(Int, nrow(ds))
 

src/byrow/util.jl

@@ -145,14 +145,14 @@ end
 
 
 function write_vals!(a, pos, x::AbstractString)
-    needed_space = length(x)
+    needed_space = ncodeunits(x)
     available_space = length(a)-pos+1
     needed_space > available_space && throw(ArgumentError("not enough space in buffer to write value into it"))
     z = Base.CodeUnits(x)
-    for i in 1:length(x)
+    for i in 1:ncodeunits(x)
         a[i+pos-1] =z[i]
     end
-    pos+length(x)
+    pos+ncodeunits(x)
 end
 
 

src/dataset/modify.jl

@@ -254,7 +254,7 @@ end
 function normalize_modify!(outidx::Index, idx, arg::AbstractVector)
     res = Any[]
     for i in 1:length(arg)
-        _res = normalize_modify!(outidx::Index, idx::Index, arg[i])
+        _res = normalize_modify!(outidx::Index, idx, arg[i])
         if _res isa AbstractVector
             for j in 1:length(_res)
                 push!(res, _res[j])

test/byrow.jl

@@ -335,6 +335,12 @@
 
     ds = Dataset(x = [1,10], x2 = ["A", "BC"], x3 = [2.0,4.54])
     @test byrow(ds, join, :, delim = "--", last = "-") == ["1--A-2.0", "10--BC-4.54"]
+
+
+    ds = Dataset(x = ["α1", "β∘1"], y = [1,2], z = [1.4,5.6])
+    @test byrow(ds, join, :) == ["α111.4", "β∘125.6"]
+    @test byrow(ds, join, :, delim = ",") == ["α1,1,1.4", "β∘1,2,5.6"]
+    @test byrow(ds, join, :, delim = ":×:") == ["α1:×:1:×:1.4", "β∘1:×:2:×:5.6"]
 end
 
 @testset "cum*/! - sort/!" begin