add support for getproperty broadcasting (#2655)

NEWS.md

@@ -26,12 +26,23 @@
   additional column to be added in the last position in the resulting data frame
   that will identify the source data frame.
   ([#2649](/~https://github.com/JuliaData/DataFrames.jl/pull/2649))
+* since Julia 1.7 using broadcasting assignment on a `DataFrame` column
+  selected as a property (e.g. `df.col .= 1`) is allowed when column does not
+  exist and it allocates a fresh column
+  ([#2655](/~https://github.com/JuliaData/DataFrames.jl/pull/2655))
 
 ## Deprecated
 
 * in `leftjoin`, `rightjoin`, and `outerjoin` the `indicator` keyword argument
   is deprecated in favor of `source` keyword argument; `indicator` will be removed
   in 2.0 release ([2649](/~https://github.com/JuliaData/DataFrames.jl/pull/2649))
+* Using broadcasting assignment on a `SubDataFrames` column selected as a property
+  (e.g. `sdf.col .= 1`) is deprecated; it will be disallowed in the future.
+  ([#2655](/~https://github.com/JuliaData/DataFrames.jl/pull/2655))
+* Broadcasting assignment to an existing column of a `DataFrame`
+  selected as a property (e.g. `df.col .= 1`) being an in-place
+  operation is deprecated. It will allocate a fresh column in the future
+  ([#2655](/~https://github.com/JuliaData/DataFrames.jl/pull/2655))
 * all deprecations present in 0.22 release now throw an error
   ([#2554](/~https://github.com/JuliaData/DataFrames.jl/pull/2554))
 

docs/src/lib/indexing.md

@@ -183,19 +183,24 @@ In such an operation `AbstractDataFrame` is considered as two-dimensional and `D
     `DataFrameRow` is considered to be column-oriented.
 
 Additional rules:
-* in the `df[CartesianIndex(row, col)] .= v`, `df[row, col] .= v` syntaxes `v` is broadcasted into the contents of `df[row, col]` (this is consistent with Julia Base);
+* in the `df[CartesianIndex(row, col)] .= v`, `df[row, col] .= v` syntaxes `v` is
+  broadcasted into the contents of `df[row, col]` (this is consistent with Julia Base);
 * in the `df[row, cols] .= v` syntaxes the assignment to `df` is performed in-place;
-* in the `df[rows, col] .= v` and `df[rows, cols] .= v` syntaxes the assignment to `df` is performed in-place;
-  if `rows` is `:` and `col` is `Symbol` or `AbstractString` and it is missing from `df` then a new column is allocated and added;
+* in the `df[rows, col] .= v` and `df[rows, cols] .= v` syntaxes the assignment to
+  `df` is performed in-place; if `rows` is `:` and `col` is `Symbol` or `AbstractString`
+  and it is missing from `df` then a new column is allocated and added;
   the length of the column is always the value of `nrow(df)` before the assignment takes place;
 * in the `df[!, col] .= v` syntax column `col` is replaced by a freshly allocated vector;
   if `col` is `Symbol` or `AbstractString` and it is missing from `df` then a new column is allocated added;
   the length of the column is always the value of `nrow(df)` before the assignment takes place;
 * the `df[!, cols] .= v` syntax replaces existing columns `cols` in data frame `df` with freshly allocated vectors;
-* `df.col .= v` syntax is allowed and performs in-place assignment to an existing vector `df.col`.
+* `df.col .= v` syntax currently performs in-place assignment to an existing vector `df.col`;
+  this behavior is deprecated and a new column will be allocated in the future.
+  Starting from Julia 1.7 if `:col` is not present in `df` then a new column will be created in `df`.
 * in the `sdf[CartesianIndex(row, col)] .= v`, `sdf[row, col] .= v` and `sdf[row, cols] .= v` syntaxes the assignment to `sdf` is performed in-place;
 * in the `sdf[rows, col] .= v` and `sdf[rows, cols] .= v` syntaxes the assignment to `sdf` is performed in-place;
-* `sdf.col .= v` syntax is allowed and performs in-place assignment to an existing vector `sdf.col`.
+* `sdf.col .= v` syntax is performs an in-place assignment to an existing vector `sdf.col` and is deprecated;
+  in the future this operation will not be allowed.
 * `dfr.col .= v` syntax is allowed and performs in-place assignment to a value extracted by `dfr.col`.
 
 Note that `sdf[!, col] .= v` and `sdf[!, cols] .= v` syntaxes are not allowed as `sdf` can be only modified in-place.

src/other/broadcasting.jl

@@ -121,6 +121,28 @@ end
 Base.dotview(df::SubDataFrame, ::typeof(!), idxs) =
     throw(ArgumentError("broadcasting with ! row selector is not allowed for SubDataFrame"))
 
+
+# TODO: remove the deprecations when Julia 1.7 functionality is commonly used
+#       by the community
+if isdefined(Base, :dotgetproperty)
+    function Base.dotgetproperty(df::DataFrame, col::SymbolOrString)
+        if columnindex(df, col) == 0
+            return LazyNewColDataFrame(df, Symbol(col))
+        else
+            Base.depwarn("In the future this operation will allocate a new column" *
+                    "instead of performing an in-place assignment.", :dotgetproperty)
+            return getproperty(df, col)
+        end
+    end
+
+    function Base.dotgetproperty(df::SubDataFrame, col::SymbolOrString)
+        Base.depwarn("broadcasting getproperty is deprecated for SubDataFrame and " *
+                     "will be disallowed in  the future. Use `df[:, $(repr(col))] .= ... instead",
+                     :dotgetproperty)
+        return getproperty(df, col)
+    end
+end
+
 function Base.copyto!(lazydf::LazyNewColDataFrame, bc::Base.Broadcast.Broadcasted{T}) where T
     if bc isa Base.Broadcast.Broadcasted{<:Base.Broadcast.AbstractArrayStyle{0}}
         bc_tmp = Base.Broadcast.Broadcasted{T}(bc.f, bc.args, ())

test/broadcasting.jl

@@ -1458,8 +1458,13 @@ end
     @test v1 == [100.0, 100.0, 100.0]
 
     df = copy(refdf)
-    @test_throws ArgumentError df.newcol .= 'd'
-    @test df == refdf
+    if isdefined(Base, :dotgetproperty)
+        df.newcol .= 'd'
+        @test df == [refdf DataFrame(newcol=fill('d', 3))]
+    else
+        @test_throws ArgumentError df.newcol .= 'd'
+        @test df == refdf
+    end
 
     df = view(copy(refdf), :, :)
     v1 = df[!, 1]
@@ -1842,4 +1847,18 @@ end
     @test_throws DimensionMismatch df[:, "z"] .= z
 end
 
+@testset "broadcasting of getproperty" begin
+    if isdefined(Base, :dotgetproperty)
+        df = DataFrame(a=1:4)
+        df.b .= 1
+        df.c .= 4:-1:1
+        # TODO: enable this in the future when the deprecation period is finished
+        # df.a .= 'a':'d'
+        # @test df.a isa Vector{Char}
+        # @test df == DataFrame(a='a':'d', b=1, c=4:-1:1)
+        # dfv = view(df, 2:3, 2:3)
+        # @test_throws ArgumentError dfv.b .= 0
+    end
+end
+
 end # module

test/deprecated.jl

@@ -10,6 +10,15 @@ const ≅ = isequal
     @test_throws ArgumentError aggregate()
 end
 
+@testset "deprecated broadcasting assignment" begin
+    df = DataFrame(a=1:4, b=1, c=2)
+    df.a .= 'a':'d'
+    @test df == DataFrame(a=97:100, b=1, c=2)
+    dfv = view(df, 2:3, 2:3)
+    dfv.b .= 0
+    @test df.b == [1, 0, 0, 1]
+end
+
 @testset "All indexing" begin
     df = DataFrame(a=1, b=2, c=3)