Merge pull request #47 from JuliaML/ox/neodataframes

Update for DataFrameRow changes

REQUIRE

@@ -3,4 +3,4 @@ StatsBase 0.13
 LearnBase 0.2.0 0.3.0
 MLLabelUtils 0.4.0
 MLDataPattern 0.5.0
-DataFrames 0.11
+DataFrames 0.17

src/datapattern.jl

@@ -6,16 +6,27 @@ import DataFrames: DataFrames, AbstractDataFrame, SubDataFrame
 LearnBase.nobs(dt::AbstractDataFrame) = DataFrames.nrow(dt)
 LearnBase.getobs(dt::AbstractDataFrame, idx) = dt[idx,:]
 
+LearnBase.nobs(dt::DataFrameRow) = 1  # it is a observation
+function LearnBase.getobs(dt::DataFrameRow, idx)
+    idx == 1:1 || throw(ArgumentError(
+         "Attempting to read multiple rows ($idx) with a single row"))
+
+    return dt
+end
+
 # custom data subset in form of SubDataFrame
 LearnBase.datasubset(dt::AbstractDataFrame, idx, ::ObsDim.Undefined) =
-    view(dt, idx)
+    @view dt[idx, :]
 
 # throw error if no target extraction function is supplied
 LearnBase.gettarget(::typeof(identity), dt::AbstractDataFrame) =
     _throw_table_error()
+LearnBase.gettarget(::typeof(identity), dt::DataFrameRow) =
+    _throw_table_error()
 
 # convenience syntax to allow column name
 LearnBase.gettarget(col::Symbol, dt::AbstractDataFrame) = dt[1, col]
+LearnBase.gettarget(col::Symbol, dt::DataFrameRow) = dt[col]
 LearnBase.gettarget(fun, dt::AbstractDataFrame) = fun(dt)
 
 # avoid copy when target extraction function is supplied

test/tst_datapattern.jl

@@ -2,20 +2,24 @@
     df = DataFrame(x1 = rand(5), x2 = rand(5), y = [:a,:a,:b,:a,:b])
 
     @test_throws ArgumentError targets(df)
+    @test_throws ArgumentError targets(df[1,:])  # DataFrameRow
 
     @test targets(:y, df) == [:a,:a,:b,:a,:b]
-    @test targets(row->row[1,:y], df) == [:a,:a,:b,:a,:b]
+    @test targets(row->row[:y], df) == [:a,:a,:b,:a,:b]
     @test typeof(targets(:y, df)) == Vector{Symbol}
-    @test eltype(targets(x->x, df)) <: SubDataFrame
 
     @test nobs(@inferred(undersample(:y, df))) === 4
 
     @test @inferred(getobs(df, 2)) == df[2,:]
     @test @inferred(getobs(datasubset(df, 2))) == df[2,:]
+    @test @inferred(getobs(df, 2:3)) == df[2:3,:]
+    @test @inferred(getobs(datasubset(df, 2:3))) == df[2:3,:]
 
-    @test typeof(datasubset(df, 2)) <: SubDataFrame
-    @test @inferred(datasubset(df, [1,2,3,5])) == view(df, [1,2,3,5])
-    @test @inferred(datasubset(df, 2)) == view(df, 2)
-    @test @inferred(datasubset(datasubset(df, 2:3), 2)) == view(df, 3)
-    @test @inferred(datasubset(df)) == view(df, 1:5)
+
+    @test typeof(datasubset(df, 2)) <: DataFrameRow
+    @test typeof(datasubset(df, 2:5)) <: SubDataFrame
+    @test @inferred(datasubset(df, [1,2,3,5])) == df[[1,2,3,5], :]
+    @test @inferred(datasubset(df, 2)) == df[2, :]
+    @test @inferred(datasubset(datasubset(df, 2:3), 2)) == df[3, :]
+    @test @inferred(datasubset(df)) == df
 end