schema interface (#39)

* `schema` interface

* throw any input with LightInterface anyway

* add tests

Author: aviatesk

src/data_utils.jl

@@ -123,6 +123,23 @@ classes(x) = classes(get_interface_mode(), x)
 
 classes(::LightInterface, x) = errlight("classes")
 
+# ------------------------------------------------------------------------
+# schema
+
+"""
+    schema(X)
+
+If `X` is tabular, inspect the column types and scitypes, otherwise return
+`nothing`.
+"""
+schema(X; kw...) = schema(get_interface_mode(), vtrait(X), X; kw...)
+
+schema(::Mode, ::Val{:other}, X; kw...) = nothing
+
+schema(::LightInterface, ::Val{:other}, X; kw...) = errlight("schema")
+
+schema(::LightInterface, ::Val{:table}, X; kw...) = errlight("schema")
+
 # ------------------------------------------------------------------------
 # decoder
 

test/data_utils.jl

@@ -58,6 +58,28 @@ end
     @test classes(x[1]) == ['a', 'b']
 end
 # ------------------------------------------------------------------------
+@testset "schema-light" begin
+    # throw error for any input anyway
+    setlight()
+    ary = rand(10, 3)
+    @test_throws M.InterfaceError M.schema(ary)
+    df = DataFrame(rand(10, 3))
+    @test_throws M.InterfaceError M.schema(df)
+end
+@testset "schema-full" begin
+    setfull()
+    ary = rand(10, 3)
+    @test M.schema(ary) === nothing
+    M.schema(::FI, ::Val{:table}, X; kw...) = schema(X; kw...) # this would be defined in MLJBase.jl
+    df = DataFrame(A = rand(10), B = categorical(rand('a':'c', 10)))
+    sch = M.schema(df)
+    @test sch.names == (:A, :B)
+    @test sch.types[1] <: Float64
+    @test sch.types[2] <: CategoricalValue
+    @test sch.scitypes[1] <: Continuous
+    @test sch.scitypes[2] <: Multiclass
+end
+# ------------------------------------------------------------------------
 @testset "decoder-light" begin
     setlight()
     x = 5