fix issues with byrow(join)

Author: sl-solution

docs/src/man/byrow.md

@@ -57,7 +57,7 @@ Generally, `byrow` is efficient for any `fun` which returns a single value for e
 * `isequal` : Return `true` when all values are equal. Optionally, a vector of values can be passed via the `with` keyword to compare equality with it.
 * `isless` : Return `true` when all values are less than passed vector(or column specified by its name) as `with`. Passing `rev = true` change `less` to `greater`.
 * `issorted` : Check if the values are sorted
-* `join`: Convert values in each row to string and join them into a single string, inserting the given delimiter (if any) between adjacent values. If `last` is given, it will be used after the last string.
+* `join`: Convert values in each row to string and join them into a single string, inserting the given delimiter (if any) between adjacent values. If `last` is given, it will be used instead of `delim` between the last two strings.
 * `maximum` : Return the maximum value
 * `mean` : Compute the mean value
 * `minimum` : Return the minimum value

src/byrow/row_functions.jl

@@ -1115,11 +1115,11 @@ row_hash(ds::AbstractDataset, cols = :; threads = true) = row_hash(ds, identity,
 function _convert_uint8_to_string!(res, init0, curr_pos, ds, threads)
     if threads
         Threads.@threads for i in 1:nrow(ds)
-            res[i] = String(view(init0, 1:curr_pos[i]-1, i))
+            res[i] = String(view(init0, 1:curr_pos[i]-2, i))
         end
     else
         for i in 1:nrow(ds)
-            res[i] = String(view(init0, 1:curr_pos[i]-1, i))
+            res[i] = String(view(init0, 1:curr_pos[i]-2, i))
         end
     end
 end
@@ -1146,14 +1146,15 @@ function row_join(ds::AbstractDataset, cols = :; threads = true, delim::Abstract
     curr_pos = ones(Int, nrow(ds))
 
     delimiter = Base.CodeUnits(delim)
-    row_join!(init0, curr_pos, ds, repeat([identity], length(colsidx)), colsidx; delim = delimiter, quotechar = nothing, threads = threads)
+    row_join!(init0, curr_pos, ds, repeat([identity], length(colsidx)-1), view(colsidx, 1:length(colsidx)-1); delim = delimiter, quotechar = nothing, threads = threads)
     if length(last)>0
-        last_uint = Base.CodeUnits(last)
-        last_len = length(last_uint)
-        _add_last_for_join!(init0, curr_pos, ds, last_uint, last_len, threads)
-    else
-        curr_pos .-= 1
+        delimiter = Base.CodeUnits(last)
+    end
+    if length(colsidx) > 1
+        _add_last_for_join!(init0, curr_pos, ds, delimiter, length(delimiter), threads)
     end
+    row_join!(init0, curr_pos, ds, [identity], colsidx[length(colsidx)]; delim = delimiter, quotechar = nothing, threads = threads)
+
     res = Vector{Union{String, Missing}}(undef, nrow(ds))
     _convert_uint8_to_string!(res, init0, curr_pos, ds, threads)
     res

test/byrow.jl

@@ -319,19 +319,22 @@
     ds = Dataset(x1 = ["A", "B,"], x2 =["TEA", "TOOOOL"])
     @test byrow(ds, join, r"x") == ["ATEA", "B,TOOOOL"]
     @test byrow(ds, join, r"x", delim = ",") == ["A,TEA", "B,,TOOOOL"]
-    @test byrow(ds, join, r"x", last = ".") == ["ATEA.", "B,TOOOOL."]
-    @test byrow(ds, join, r"x", last = "end", delim = "/-/") == ["A/-/TEAend", "B,/-/TOOOOLend"]
+    @test byrow(ds, join, r"x", last = ".") == ["A.TEA", "B,.TOOOOL"]
+    @test byrow(ds, join, r"x", last = "end", delim = "/-/") == ["AendTEA", "B,endTOOOOL"]
 
     @test byrow(view(ds, [1,2], :), join, r"x") == ["ATEA", "B,TOOOOL"]
     @test byrow(view(ds, [1,2], :), join, r"x", delim = ",") == ["A,TEA", "B,,TOOOOL"]
-    @test byrow(view(ds, [1,2], :), join, r"x", last = ".") == ["ATEA.", "B,TOOOOL."]
-    @test byrow(view(ds, [1,2], :), join, r"x", last = "end", delim = "/-/") == ["A/-/TEAend", "B,/-/TOOOOLend"]
+    @test byrow(view(ds, [1,2], :), join, r"x", last = ".") == ["A.TEA", "B,.TOOOOL"]
+    @test byrow(view(ds, [1,2], :), join, r"x", last = "end", delim = "/-/") == ["AendTEA", "B,endTOOOOL"]
 
     repeat!(ds, 1000)
     @test byrow(ds, join, r"x", threads = true) == repeat(["ATEA", "B,TOOOOL"], 1000)
     @test byrow(ds, join, r"x", threads = true, delim = ",") == repeat(["A,TEA", "B,,TOOOOL"], 1000)
-    @test byrow(ds, join, r"x", threads = true, last = ".") == repeat(["ATEA.", "B,TOOOOL."], 1000)
-    @test byrow(ds, join, r"x", threads = true, last = "end", delim = "/-/") == repeat(["A/-/TEAend", "B,/-/TOOOOLend"], 1000)
+    @test byrow(ds, join, r"x", threads = true, last = ".") == repeat(["A.TEA", "B,.TOOOOL"], 1000)
+    @test byrow(ds, join, r"x", threads = true, last = "end", delim = "/-/") == repeat(["AendTEA", "B,endTOOOOL"], 1000)
+
+    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"]
 end
 
 @testset "cum*/! - sort/!" begin