Say good-bye to transpose ISchema

Author: wschin

src/Microsoft.ML.Data/Commands/ShowSchemaCommand.cs

@@ -91,11 +91,11 @@ internal static void RunOnData(TextWriter writer, Arguments args, IDataView data
                 foreach (var view in viewChainReversed.Reverse())
                 {
                     writer.WriteLine("---- {0} ----", view.GetType().Name);
-                    PrintSchema(writer, args, view.Schema, (view as ITransposeDataView)?.TransposeSchema);
+                    PrintSchema(writer, args, view.Schema, (view as ITransposeDataView)?.TransposeSlotTypeHolder);
                 }
             }
             else
-                PrintSchema(writer, args, data.Schema, (data as ITransposeDataView)?.TransposeSchema);
+                PrintSchema(writer, args, data.Schema, (data as ITransposeDataView)?.TransposeSlotTypeHolder);
         }
 
         /// <summary>
@@ -113,12 +113,12 @@ private static IEnumerable<IDataView> GetViewChainReversed(IDataView data)
             }
         }
 
-        private static void PrintSchema(TextWriter writer, Arguments args, Schema schema, ITransposeSchema tschema)
+        private static void PrintSchema(TextWriter writer, Arguments args, Schema schema, ITransposeSlotTypeHolder transposeSlotTypeHolder)
         {
             Contracts.AssertValue(writer);
             Contracts.AssertValue(args);
             Contracts.AssertValue(schema);
-            Contracts.AssertValueOrNull(tschema);
+            Contracts.AssertValueOrNull(transposeSlotTypeHolder);
 #if !CORECLR
             if (args.ShowJson)
             {
@@ -137,7 +137,7 @@ private static void PrintSchema(TextWriter writer, Arguments args, Schema schema
                 {
                     var name = schema[col].Name;
                     var type = schema[col].Type;
-                    var slotType = tschema == null ? null : tschema.GetSlotType(col);
+                    var slotType = transposeSlotTypeHolder == null ? null : transposeSlotTypeHolder.GetSlotType(col);
                     itw.WriteLine("{0}: {1}{2}", name, type, slotType == null ? "" : " (T)");
 
                     bool metaVals = args.ShowMetadataValues;

src/Microsoft.ML.Data/Data/ITransposeDataView.cs

@@ -20,18 +20,17 @@ namespace Microsoft.ML.Data
     /// <see cref="IDataView"/>.
     ///
     /// The interface only advertises that columns may be accessible in slot-wise fashion. A column
-    /// is accessible in this fashion iff <see cref="TransposeSchema"/>'s
-    /// <see cref="ITransposeSchema.GetSlotType"/> returns a non-null value.
+    /// is accessible in this fashion iff <see cref="TransposeSlotTypeHolder"/>'s
+    /// <see cref="ITransposeSlotTypeHolder.GetSlotType"/> returns a non-null value.
     /// </summary>
     [BestFriend]
     internal interface ITransposeDataView : IDataView
     {
         /// <summary>
         /// An enhanced schema, containing information on the transposition properties, if any,
-        /// of each column. Note that there is no contract or suggestion that this property
-        /// should be equal to <see cref="IDataView.Schema"/>.
+        /// of each column.
         /// </summary>
-        ITransposeSchema TransposeSchema { get; }
+        ITransposeSlotTypeHolder TransposeSlotTypeHolder { get; }
 
         /// <summary>
         /// Presents a cursor over the slots of a transposable column, or throws if the column
@@ -44,15 +43,15 @@ internal interface ITransposeDataView : IDataView
     /// The transpose schema returns the schema information of the view we have transposed.
     /// </summary>
     [BestFriend]
-    internal interface ITransposeSchema : ISchema
+    internal interface ITransposeSlotTypeHolder
     {
         /// <summary>
-        /// Analogous to <see cref="ISchema.GetColumnType"/>, except instead of returning the type of value
-        /// accessible through the <see cref="RowCursor"/>, returns the item type of value accessible
-        /// through the <see cref="SlotCursor"/>. This will return <c>null</c> iff this particular
-        /// column is not transposable, that is, it cannot be viewed in a slotwise fashion. Observe from
-        /// the return type that this will always be a vector type. This vector type should be of fixed
-        /// size and one dimension.
+        /// Analogous to <see cref="ColumnType"/> in <see cref="Schema"/>, except instead of returning
+        /// the type of value accessible through the <see cref="RowCursor"/>, returns the item type of
+        /// value accessible through the <see cref="SlotCursor"/>. This will return <c>null</c> iff
+        /// this particular column is not transposable, that is, it cannot be viewed in a slotwise fashion.
+        /// Observe from the return type that this will always be a vector type. This vector type should
+        /// be of fixed size and one dimension.
         /// </summary>
         VectorType GetSlotType(int col);
     }

src/Microsoft.ML.Data/Data/SlotCursor.cs

@@ -45,14 +45,14 @@ private protected SlotCursor(IChannelProvider provider)
 
         /// <summary>
         /// The slot type for this cursor. Note that this should equal the
-        /// <see cref="ITransposeSchema.GetSlotType"/> for the column from which this slot cursor
+        /// <see cref="ITransposeSlotTypeHolder.GetSlotType"/> for the column from which this slot cursor
         /// was created.
         /// </summary>
         public abstract VectorType GetSlotType();
 
         /// <summary>
         /// A getter delegate for the slot values. The type <typeparamref name="TValue"/> must correspond
-        /// to the item type from <see cref="ITransposeSchema.GetSlotType"/>.
+        /// to the item type from <see cref="ITransposeSlotTypeHolder.GetSlotType"/>.
         /// </summary>
         public abstract ValueGetter<VBuffer<TValue>> GetGetter<TValue>();
 

src/Microsoft.ML.Data/DataLoadSave/CompositeDataLoader.cs

@@ -408,7 +408,7 @@ private CompositeDataLoader(IHost host, TransformEx[] transforms)
 
             View = transforms[transforms.Length - 1].Transform;
             _tview = View as ITransposeDataView;
-            _transposeSchema = _tview?.TransposeSchema ?? new TransposerUtils.SimpleTransposeSchema(View.Schema);
+            _transposeSlotTypeHolder = _tview?.TransposeSlotTypeHolder ?? new TransposerUtils.SimpleTransposeSchema(View.Schema);
 
             var srcLoader = transforms[0].Transform.Source as IDataLoader;
 
@@ -565,8 +565,8 @@ private static string GenerateTag(int index)
 
         public Schema Schema => View.Schema;
 
-        private readonly ITransposeSchema _transposeSchema;
-        ITransposeSchema ITransposeDataView.TransposeSchema => _transposeSchema;
+        private readonly ITransposeSlotTypeHolder _transposeSlotTypeHolder;
+        ITransposeSlotTypeHolder ITransposeDataView.TransposeSlotTypeHolder => _transposeSlotTypeHolder;
 
         public RowCursor GetRowCursor(Func<int, bool> predicate, Random rand = null)
         {
@@ -585,7 +585,7 @@ public RowCursor[] GetRowCursorSet(Func<int, bool> predicate, int n, Random rand
         public SlotCursor GetSlotCursor(int col)
         {
             _host.CheckParam(0 <= col && col < Schema.Count, nameof(col));
-            if (_transposeSchema?.GetSlotType(col) == null)
+            if (_transposeSlotTypeHolder?.GetSlotType(col) == null)
             {
                 throw _host.ExceptParam(nameof(col), "Bad call to GetSlotCursor on untransposable column '{0}'",
                     Schema[col].Name);

src/Microsoft.ML.Data/DataLoadSave/Transpose/TransposeLoader.cs

@@ -320,7 +320,7 @@ protected override void VerifyView(IDataView view)
         private readonly IMultiStreamSource _file;
         private readonly IHost _host;
         private readonly Header _header;
-        private readonly SchemaImpl _schema;
+        private readonly TransposeSlotTypeHolder _transposeSlotTypeHolder;
 
         // This is a sub-IDV holding the schema, and optionally the data stored in row-wise format.
         private readonly SubIdvEntry.SchemaSubIdv _schemaEntry;
@@ -364,9 +364,9 @@ private static VersionInfo GetVersionInfo()
         // inspect the schema. We also want to ensure that the useful property that
         // a cursor and view's schemas are the same, is preserved, which allows us
         // to use the cursors from the schema view if convenient to do so.
-        public Schema Schema { get { return _schemaEntry.GetView().Schema; } }
+        public Schema Schema => _schemaEntry.GetView().Schema;
 
-        ITransposeSchema ITransposeDataView.TransposeSchema { get { return _schema; } }
+        ITransposeSlotTypeHolder ITransposeDataView.TransposeSlotTypeHolder => _transposeSlotTypeHolder;
 
         /// <summary>
         /// Whether the master schema sub-IDV has the actual data.
@@ -411,7 +411,7 @@ public TransposeLoader(IHostEnvironment env, Arguments args, IMultiStreamSource
                 _entries = new SubIdvEntry.TransposedSubIdv[_header.ColumnCount];
                 for (int c = 0; c < _entries.Length; ++c)
                     _entries[c] = new SubIdvEntry.TransposedSubIdv(this, reader, c);
-                _schema = new SchemaImpl(this);
+                _transposeSlotTypeHolder = new TransposeSlotTypeHolder(this);
                 if (!HasRowData)
                 {
                     _colTransposers = new Transposer[_header.ColumnCount];
@@ -445,7 +445,7 @@ private TransposeLoader(IHost host, ModelLoadContext ctx, IMultiStreamSource fil
                 _entries = new SubIdvEntry.TransposedSubIdv[_header.ColumnCount];
                 for (int c = 0; c < _entries.Length; ++c)
                     _entries[c] = new SubIdvEntry.TransposedSubIdv(this, reader, c);
-                _schema = new SchemaImpl(this);
+                _transposeSlotTypeHolder = new TransposeSlotTypeHolder(this);
                 if (!HasRowData)
                 {
                     _colTransposers = new Transposer[_header.ColumnCount];
@@ -608,50 +608,20 @@ private unsafe Header InitHeader(BinaryReader reader)
             return header;
         }
 
-        private sealed class SchemaImpl : ITransposeSchema
+        private sealed class TransposeSlotTypeHolder : ITransposeSlotTypeHolder
         {
             private readonly TransposeLoader _parent;
             private Schema Schema { get { return _parent.Schema; } }
             private IHost Host { get { return _parent._host; } }
             public int ColumnCount { get { return Schema.Count; } }
 
-            public SchemaImpl(TransposeLoader parent)
+            public TransposeSlotTypeHolder(TransposeLoader parent)
             {
                 Contracts.AssertValue(parent);
                 _parent = parent;
                 var view = parent._schemaEntry.GetView().Schema;
             }
 
-            public string GetColumnName(int col)
-            {
-                return Schema[col].Name;
-            }
-
-            public bool TryGetColumnIndex(string name, out int col)
-            {
-                return Schema.TryGetColumnIndex(name, out col);
-            }
-
-            public ColumnType GetColumnType(int col)
-            {
-                return Schema[col].Type;
-            }
-
-            public ColumnType GetMetadataTypeOrNull(string kind, int col)
-            {
-                return Schema[col].Metadata.Schema.GetColumnOrNull(kind)?.Type;
-            }
-
-            public IEnumerable<KeyValuePair<string, ColumnType>> GetMetadataTypes(int col)
-            {
-                return Schema[col].Metadata.Schema.Select(c => new KeyValuePair<string, ColumnType>(c.Name, c.Type));
-            }
-
-            public void GetMetadata<TValue>(string kind, int col, ref TValue value)
-            {
-                Schema[col].Metadata.GetValue(kind, ref value);
-            }
-
             public VectorType GetSlotType(int col)
             {
                 Host.CheckParam(0 <= col && col < ColumnCount, nameof(col));
@@ -696,7 +666,7 @@ public SlotCursor GetSlotCursor(int col)
             _host.CheckParam(0 <= col && col < _header.ColumnCount, nameof(col));
             // We don't want the type error, if there is one, to be handled by the get-getter, because
             // at the point we've gotten the interior cursor, but not yet constructed the slot cursor.
-            ColumnType cursorType = _schema.GetSlotType(col).ItemType;
+            ColumnType cursorType = _transposeSlotTypeHolder.GetSlotType(col).ItemType;
             RowCursor inputCursor = view.GetRowCursor(c => true);
             try
             {
@@ -781,8 +751,8 @@ private Transposer EnsureAndGetTransposer(int col)
                         _host.AssertValue(view);
                         _host.Assert(view.Schema.Count == 1);
                         var trans = _colTransposers[col] = Transposer.Create(_host, view, false, new int[] { 0 });
-                        _host.Assert(((ITransposeDataView)trans).TransposeSchema.ColumnCount == 1);
-                        _host.Assert(((ITransposeDataView)trans).TransposeSchema.GetSlotType(0).ValueCount == Schema[col].Type.ValueCount);
+                        _host.Assert(trans.Schema.Count == 1);
+                        _host.Assert(trans.TransposeSlotTypeHolder.GetSlotType(0).ValueCount == Schema[col].Type.ValueCount);
                     }
                 }
             }
@@ -843,7 +813,7 @@ private void Init(int col)
                 Ch.Assert(0 <= col && col < Schema.Count);
                 Ch.Assert(_colToActivesIndex[col] >= 0);
                 var type = Schema[col].Type;
-                Ch.Assert(((ITransposeDataView)_parent).TransposeSchema.GetSlotType(col).ValueCount == _parent._header.RowCount);
+                Ch.Assert(((ITransposeDataView)_parent).TransposeSlotTypeHolder.GetSlotType(col).ValueCount == _parent._header.RowCount);
                 Action<int> func = InitOne<int>;
                 if (type.IsVector)
                     func = InitVec<int>;

src/Microsoft.ML.Data/DataLoadSave/Transpose/TransposeSaver.cs

@@ -109,7 +109,7 @@ private void SaveTransposedData(IChannel ch, Stream stream, ITransposeDataView d
             header.Signature = TransposeLoader.Header.SignatureValue;
             header.Version = TransposeLoader.Header.WriterVersion;
             header.CompatibleVersion = TransposeLoader.Header.WriterVersion;
-            VectorType slotType = data.TransposeSchema.GetSlotType(cols[0]);
+            VectorType slotType = data.TransposeSlotTypeHolder.GetSlotType(cols[0]);
             ch.AssertValue(slotType);
             header.RowCount = slotType.ValueCount;
             header.ColumnCount = cols.Length;