Microsoft.ML.Transforms assembly lockdown

src/Microsoft.ML.Data/Transforms/Hashing.cs

@@ -136,7 +136,7 @@ private static VersionInfo GetVersionInfo()
         private readonly VBuffer<ReadOnlyMemory<char>>[] _keyValues;
         private readonly VectorType[] _kvTypes;
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var type = inputSchema[srcCol].Type;
             if (!HashingEstimator.IsColumnTypeValid(type))

src/Microsoft.ML.Data/Transforms/KeyToVector.cs

@@ -112,7 +112,7 @@ private string TestIsKey(DataViewType type)
             return "key type of known cardinality";
         }
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var type = inputSchema[srcCol].Type;
             string reason = TestIsKey(type);

src/Microsoft.ML.Data/Transforms/Normalizer.cs

@@ -561,7 +561,7 @@ private protected override void SaveModel(ModelSaveContext ctx)
             }
         }
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             const string expectedType = "scalar or known-size vector of R4";
 

src/Microsoft.ML.Data/Transforms/OneToOneTransformerBase.cs

@@ -78,7 +78,8 @@ private void CheckInput(DataViewSchema inputSchema, int col, out int srcCol)
             CheckInputColumn(inputSchema, col, srcCol);
         }
 
-        protected virtual void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        [BestFriend]
+        private protected virtual void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             // By default, there are no extra checks.
         }

src/Microsoft.ML.HalLearners/VectorWhitening.cs

@@ -209,7 +209,7 @@ internal static IRowMapper Create(IHostEnvironment env, ModelLoadContext ctx, Da
         private static (string outputColumnName, string inputColumnName)[] GetColumnPairs(VectorWhiteningEstimator.ColumnInfo[] columns)
             => columns.Select(c => (c.Name, c.InputColumnName ?? c.Name)).ToArray();
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var inType = inputSchema[srcCol].Type;
             var reason = TestColumn(inType);

src/Microsoft.ML.ImageAnalytics/ImageGrayscale.cs

@@ -164,7 +164,7 @@ private protected override void SaveModel(ModelSaveContext ctx)
 
         private protected override IRowMapper MakeRowMapper(DataViewSchema schema) => new Mapper(this, schema);
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             if (!(inputSchema[srcCol].Type is ImageType))
                 throw Host.ExceptSchemaMismatch(nameof(inputSchema), "input", ColumnPairs[col].inputColumnName, "image", inputSchema[srcCol].Type.ToString());

src/Microsoft.ML.ImageAnalytics/ImageLoader.cs

@@ -131,7 +131,7 @@ private static IDataTransform Create(IHostEnvironment env, ModelLoadContext ctx,
         private static IRowMapper Create(IHostEnvironment env, ModelLoadContext ctx, DataViewSchema inputSchema)
             => Create(env, ctx).MakeRowMapper(inputSchema);
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             if (!(inputSchema[srcCol].Type is TextDataViewType))
                 throw Host.ExceptSchemaMismatch(nameof(inputSchema), "input", ColumnPairs[col].inputColumnName, TextDataViewType.Instance.ToString(), inputSchema[srcCol].Type.ToString());

src/Microsoft.ML.ImageAnalytics/ImagePixelExtractor.cs

@@ -272,7 +272,7 @@ private protected override void SaveModel(ModelSaveContext ctx)
 
         private protected override IRowMapper MakeRowMapper(DataViewSchema schema) => new Mapper(this, schema);
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var inputColName = _columns[col].InputColumnName;
             var imageType = inputSchema[srcCol].Type as ImageType;

src/Microsoft.ML.ImageAnalytics/ImageResizer.cs

@@ -265,7 +265,7 @@ private protected override void SaveModel(ModelSaveContext ctx)
 
         private protected override IRowMapper MakeRowMapper(DataViewSchema schema) => new Mapper(this, schema);
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             if (!(inputSchema[srcCol].Type is ImageType))
                 throw Host.ExceptSchemaMismatch(nameof(inputSchema), "input", _columns[col].InputColumnName, "image", inputSchema[srcCol].Type.ToString());

src/Microsoft.ML.PCA/PcaTransformer.cs

@@ -499,7 +499,7 @@ private float[][] PostProcess(float[][] y, float[] sigma, float[] z, int d, int
 
         private protected override IRowMapper MakeRowMapper(DataViewSchema schema) => new Mapper(this, schema);
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             ValidatePcaInput(Host, inputSchema[srcCol].Name, inputSchema[srcCol].Type);
         }

src/Microsoft.ML.Transforms/CategoricalCatalog.cs

@@ -9,7 +9,7 @@
 namespace Microsoft.ML
 {
     /// <summary>
-    /// Static extensions for categorical transforms.
+    /// The catalog of categorical transformations.
     /// </summary>
     public static class CategoricalCatalog
     {

src/Microsoft.ML.Transforms/CompositeTransformer.cs

@@ -16,7 +16,7 @@
 
 namespace Microsoft.ML.Transforms
 {
-    public static class CompositeTransformer
+    internal static class CompositeTransformer
     {
         private const string RegistrationName = "CompositeTransform";
 

src/Microsoft.ML.Transforms/FourierDistributionSampler.cs

@@ -30,15 +30,16 @@ namespace Microsoft.ML.Transforms
     /// <summary>
     /// Signature for an IFourierDistributionSampler constructor.
     /// </summary>
-    public delegate void SignatureFourierDistributionSampler(float avgDist);
+    [BestFriend]
+    internal delegate void SignatureFourierDistributionSampler(float avgDist);
 
     public interface IFourierDistributionSampler : ICanSaveModel
     {
         float Next(Random rand);
     }
 
     [TlcModule.ComponentKind("FourierDistributionSampler")]
-    public interface IFourierDistributionSamplerFactory : IComponentFactory<float, IFourierDistributionSampler>
+    internal interface IFourierDistributionSamplerFactory : IComponentFactory<float, IFourierDistributionSampler>
     {
     }
 
@@ -51,10 +52,11 @@ public sealed class Options : IFourierDistributionSamplerFactory
             [Argument(ArgumentType.AtMostOnce, HelpText = "gamma in the kernel definition: exp(-gamma*||x-y||^2 / r^2). r is an estimate of the average intra-example distance", ShortName = "g")]
             public float Gamma = 1;
 
-            public IFourierDistributionSampler CreateComponent(IHostEnvironment env, float avgDist) => new GaussianFourierSampler(env, this, avgDist);
+            IFourierDistributionSampler IComponentFactory<float, IFourierDistributionSampler>.CreateComponent(IHostEnvironment env, float avgDist)
+                => new GaussianFourierSampler(env, this, avgDist);
         }
 
-        public const string LoaderSignature = "RandGaussFourierExec";
+        internal const string LoaderSignature = "RandGaussFourierExec";
         private static VersionInfo GetVersionInfo()
         {
             return new VersionInfo(
@@ -66,7 +68,7 @@ private static VersionInfo GetVersionInfo()
                 loaderAssemblyName: typeof(GaussianFourierSampler).Assembly.FullName);
         }
 
-        public const string LoadName = "GaussianRandom";
+        internal const string LoadName = "GaussianRandom";
 
         private readonly float _gamma;
 
@@ -79,7 +81,7 @@ public GaussianFourierSampler(IHostEnvironment env, Options args, float avgDist)
             _gamma = args.Gamma / avgDist;
         }
 
-        public static GaussianFourierSampler Create(IHostEnvironment env, ModelLoadContext ctx)
+        private static GaussianFourierSampler Create(IHostEnvironment env, ModelLoadContext ctx)
         {
             Contracts.CheckValue(env, nameof(env));
             env.CheckValue(ctx, nameof(ctx));
@@ -130,7 +132,8 @@ public sealed class Options : IFourierDistributionSamplerFactory
             [Argument(ArgumentType.AtMostOnce, HelpText = "a in the term exp(-a|x| / r). r is an estimate of the average intra-example L1 distance")]
             public float A = 1;
 
-            public IFourierDistributionSampler CreateComponent(IHostEnvironment env, float avgDist) => new LaplacianFourierSampler(env, this, avgDist);
+            IFourierDistributionSampler IComponentFactory<float, IFourierDistributionSampler>.CreateComponent(IHostEnvironment env, float avgDist)
+                => new LaplacianFourierSampler(env, this, avgDist);
         }
 
         private static VersionInfo GetVersionInfo()
@@ -144,8 +147,8 @@ private static VersionInfo GetVersionInfo()
                 loaderAssemblyName: typeof(LaplacianFourierSampler).Assembly.FullName);
         }
 
-        public const string LoaderSignature = "RandLaplacianFourierExec";
-        public const string RegistrationName = "LaplacianRandom";
+        internal const string LoaderSignature = "RandLaplacianFourierExec";
+        internal const string RegistrationName = "LaplacianRandom";
 
         private readonly IHost _host;
         private readonly float _a;
@@ -159,7 +162,7 @@ public LaplacianFourierSampler(IHostEnvironment env, Options args, float avgDist
             _a = args.A / avgDist;
         }
 
-        public static LaplacianFourierSampler Create(IHostEnvironment env, ModelLoadContext ctx)
+        private static LaplacianFourierSampler Create(IHostEnvironment env, ModelLoadContext ctx)
         {
             Contracts.CheckValue(env, nameof(env));
             env.CheckValue(ctx, nameof(ctx));

src/Microsoft.ML.Transforms/GcnTransform.cs

@@ -202,7 +202,7 @@ private static (string outputColumnName, string inputColumnName)[] GetColumnPair
             return columns.Select(x => (x.Name, x.InputColumnName)).ToArray();
         }
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var inType = inputSchema[srcCol].Type;
             if (!LpNormalizingEstimatorBase.IsColumnTypeValid(inType))

src/Microsoft.ML.Transforms/KeyToVectorMapping.cs

@@ -69,7 +69,7 @@ private string TestIsKey(DataViewType type)
             return "key type of known cardinality";
         }
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var type = inputSchema[srcCol].Type;
             string reason = TestIsKey(type);

src/Microsoft.ML.Transforms/MissingValueDroppingTransformer.cs

@@ -103,7 +103,7 @@ private MissingValueDroppingTransformer(IHostEnvironment env, ModelLoadContext c
         private static (string outputColumnName, string inputColumnName)[] GetColumnPairs(Column[] columns)
             => columns.Select(c => (c.Name, c.Source ?? c.Name)).ToArray();
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var inType = inputSchema[srcCol].Type;
             if (!(inType is VectorType))

src/Microsoft.ML.Transforms/MissingValueReplacing.cs

@@ -194,7 +194,7 @@ private static (string outputColumnName, string inputColumnName)[] GetColumnPair
         // REVIEW: Currently these arrays are constructed on load but could be changed to being constructed lazily.
         private readonly BitArray[] _repIsDefault;
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var type = inputSchema[srcCol].Type;
             string reason = TestType(type);

src/Microsoft.ML.Transforms/ProjectionCatalog.cs

@@ -7,6 +7,9 @@
 
 namespace Microsoft.ML
 {
+    /// <summary>
+    /// The catalog of projection transformations.
+    /// </summary>
     public static class ProjectionCatalog
     {
         /// <summary>

src/Microsoft.ML.Transforms/RandomFourierFeaturizing.cs

@@ -240,7 +240,7 @@ private static (string outputColumnName, string inputColumnName)[] GetColumnPair
             return columns.Select(x => (x.Name, x.InputColumnName)).ToArray();
         }
 
-        protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
+        private protected override void CheckInputColumn(DataViewSchema inputSchema, int col, int srcCol)
         {
             var type = inputSchema[srcCol].Type;
             string reason = TestColumnType(type);

src/Microsoft.ML.Transforms/Text/NgramHashingTransformer.cs

@@ -875,15 +875,31 @@ public sealed class NgramHashingEstimator : IEstimator<NgramHashingTransformer>
         /// </summary>
         public sealed class ColumnInfo
         {
+            /// <summary>Name of the column resulting from the transformation of <see cref="InputColumnNames"/>.</summary>
             public readonly string Name;
+            /// <summary>Names of the columns to transform.</summary>
             public readonly string[] InputColumnNames;
+            /// <summary>Maximum ngram length.</summary>
             public readonly int NgramLength;
+            /// <summary>Maximum number of tokens to skip when constructing an ngram.</summary>
             public readonly int SkipLength;
+            /// <summary>Whether to store all ngram lengths up to <see cref="NgramLength"/>, or only <see cref="NgramLength"/>.</summary>
             public readonly bool AllLengths;
+            /// <summary>Number of bits to hash into. Must be between 1 and 31, inclusive.</summary>
             public readonly int HashBits;
+            /// <summary>Hashing seed.</summary>
             public readonly uint Seed;
+            /// <summary>Whether the position of each term should be included in the hash.</summary>
             public readonly bool Ordered;
+            /// <summary>
+            /// During hashing we constuct mappings between original values and the produced hash values.
+            /// Text representation of original values are stored in the slot names of the  metadata for the new column.
+            /// Hashing, as such, can map many initial values to one.
+            /// <see cref="InvertHash"/> specifies the upper bound of the number of distinct input values mapping to a hash that should be retained.
+            /// <value>0</value> does not retain any input values. <value>-1</value> retains all input values mapping to each hash.
+            /// </summary>
             public readonly int InvertHash;
+            /// <summary>Whether to rehash unigrams.</summary>
             public readonly bool RehashUnigrams;
             // For all source columns, use these friendly names for the source
             // column names instead of the real column names.
@@ -893,15 +909,16 @@ public sealed class ColumnInfo
             /// Describes how the transformer handles one column pair.
             /// </summary>
             /// <param name="name">Name of the column resulting from the transformation of <paramref name="inputColumnNames"/>.</param>
-            /// <param name="inputColumnNames">Name of the columns to transform. </param>
+            /// <param name="inputColumnNames">Names of the columns to transform. </param>
             /// <param name="ngramLength">Maximum ngram length.</param>
             /// <param name="skipLength">Maximum number of tokens to skip when constructing an ngram.</param>
-            /// <param name="allLengths">"Whether to store all ngram lengths up to ngramLength, or only ngramLength.</param>
+            /// <param name="allLengths">Whether to store all ngram lengths up to <paramref name="ngramLength"/>, or only <paramref name="ngramLength"/>.</param>
             /// <param name="hashBits">Number of bits to hash into. Must be between 1 and 31, inclusive.</param>
             /// <param name="seed">Hashing seed.</param>
             /// <param name="ordered">Whether the position of each term should be included in the hash.</param>
             /// <param name="invertHash">During hashing we constuct mappings between original values and the produced hash values.
-            /// Text representation of original values are stored in the slot names of the  metadata for the new column.Hashing, as such, can map many initial values to one.
+            /// Text representation of original values are stored in the slot names of the metadata for the new column.
+            /// Hashing, as such, can map many initial values to one.
             /// <paramref name="invertHash"/> specifies the upper bound of the number of distinct input values mapping to a hash that should be retained.
             /// <value>0</value> does not retain any input values. <value>-1</value> retains all input values mapping to each hash.</param>
             /// <param name="rehashUnigrams">Whether to rehash unigrams.</param>