Implements both Tensor and NdArray interfaces from same instance

ndarray/src/main/java/org/tensorflow/ndarray/NdArrays.java

@@ -65,7 +65,7 @@ public static ByteNdArray vectorOf(byte... values) {
     if (values == null) {
       throw new IllegalArgumentException("Values cannot be null");
     }
-    return wrap(DataBuffers.of(values, false, false), Shape.of(values.length));
+    return wrap(Shape.of(values.length), DataBuffers.of(values, false, false));
   }
 
   /**
@@ -81,19 +81,19 @@ public static ByteNdArray ofBytes(Shape shape) {
     if (shape == null) {
       throw new IllegalArgumentException("Shape cannot be null");
     }
-    return wrap(DataBuffers.ofBytes(shape.size()), shape);
+    return wrap(shape, DataBuffers.ofBytes(shape.size()));
   }
 
   /**
    * Wraps a buffer in a byte N-dimensional array of a given shape.
    *
-   * @param buffer buffer to wrap
    * @param shape shape of the array
+   * @param buffer buffer to wrap
    * @return new byte N-dimensional array
    * @throws IllegalArgumentException if shape is null, has unknown dimensions or has size bigger
    *                                  in the buffer size
    */
-  public static ByteNdArray wrap(ByteDataBuffer buffer, Shape shape) {
+  public static ByteNdArray wrap(Shape shape, ByteDataBuffer buffer) {
     return ByteDenseNdArray.create(buffer, shape);
   }
 

tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc

@@ -237,12 +237,12 @@ void RenderFactoryMethods(const OpSpec& op, const Type& op_class,
   writer->EndLine();
   for (const ArgumentSpec& input : op.inputs()) {
     if (input.iterable()) {
-      writer->Append("opBuilder.addInputList(Operands.asOutputs(" +
+      writer->Append("opBuilder.addInputList(Operands.asOutputs(scope, " +
                      input.var().name() + "));");
       writer->EndLine();
     } else {
       writer->Append("opBuilder.addInput(" + input.var().name() +
-                     ".asOutput());");
+                     ".asOutput(scope));");
       writer->EndLine();
     }
   }
@@ -348,8 +348,9 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   if (mode == OPERAND) {
     bool cast2obj = output.type().wildcard();
     Type return_type = Type::Class("Output", "org.tensorflow")
-        .add_parameter(cast2obj ? Type::Class("TType", "org.tensorflow.types.family") : output.type());
+        .add_parameter(cast2obj ? Type::Interface("TType", "org.tensorflow.types.family") : output.type());
     Method as_output = Method::Create("asOutput", return_type)
+        .add_argument(Variable::Create("scope", Type::Class("Scope", "org.tensorflow.op")))
         .add_annotation(Annotation::Create("Override"));
     if (cast2obj) {
       as_output.add_annotation(
@@ -366,7 +367,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   } else if (mode == LIST_OPERAND) {
     Type operand = Type::Interface("Operand", "org.tensorflow");
     if (output.type().wildcard()) {
-      operand.add_parameter(Type::Class("TType", "org.tensorflow.types.family"));
+      operand.add_parameter(Type::Interface("TType", "org.tensorflow.types.family"));
     } else {
       operand.add_parameter(output.type());
     }
@@ -430,10 +431,9 @@ void GenerateOp(const OpSpec& op, const EndpointSpec& endpoint,
   RenderMode mode = DEFAULT;
   if (op.outputs().size() == 1) {
     const ArgumentSpec& output = op.outputs().front();
-    Type operand_type(output.type().wildcard() ? Type::Class("TType", "org.tensorflow.types.family")
+    Type operand_type(output.type().wildcard() ? Type::Interface("TType", "org.tensorflow.types.family")
                                                : output.type());
-    Type operand_inf(Type::Interface("Operand", "org.tensorflow")
-                         .add_parameter(operand_type));
+    Type operand_inf(Type::Interface("Operand", "org.tensorflow").add_parameter(operand_type));
     if (output.iterable()) {
       mode = LIST_OPERAND;
       op_class.add_supertype(Type::IterableOf(operand_inf));

tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc

@@ -65,6 +65,13 @@ class TypeResolver {
   std::pair<Type, Type> TypesOf(const OpDef_AttrDef& attr_def,
                                 bool* iterable_out);
 
+  // Returns the highest type family this attribute is part of
+  //
+  // For example, if the attribute is of type 'bool', the base 'TType' family
+  // returned. But if it represents a number, like a float or an integer,
+  // then 'TNumber' (which supersedes 'TType') is returned.
+  Type FamilyOf(const OpDef_AttrDef& attr_def);
+
   // Returns true if the type of this attribute has already been resolved
   bool IsAttributeVisited(const string& attr_name) {
     return visited_attrs_.find(attr_name) != visited_attrs_.cend();
@@ -81,13 +88,12 @@ class TypeResolver {
   std::pair<Type, Type> MakeTypePair(const Type& type) {
     return std::make_pair(type, type);
   }
-  Type NextGeneric() {
+  Type NextGeneric(const Type& typeFamily) {
     char generic_letter = next_generic_letter_++;
     if (next_generic_letter_ > 'Z') {
       next_generic_letter_ = 'A';
     }
-    return Type::Generic(string(1, generic_letter))
-        .add_supertype(Type::Class("TType", "org.tensorflow.types.family"));
+    return Type::Generic(string(1, generic_letter)).add_supertype(typeFamily);
   }
 };
 
@@ -156,10 +162,7 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
                              .add_parameter(Type::Wildcard()));
 
   } else if (attr_type == "type") {
-    Type type = *iterable_out ? Type::Wildcard() : NextGeneric();
-    if (IsRealNumbers(attr_def.allowed_values())) {
-      type.add_supertype(Type::Class("TNumber", "org.tensorflow.types.family"));
-    }
+    Type type = *iterable_out ? Type::Wildcard() : NextGeneric(FamilyOf(attr_def));
     types = MakeTypePair(type, Type::Enum("DataType", "org.tensorflow"));
 
   } else {
@@ -170,6 +173,14 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
   return types;
 }
 
+Type TypeResolver::FamilyOf(const OpDef_AttrDef& attr_def) {
+  // TODO (karlllessard): add more type families
+  if (IsRealNumbers(attr_def.allowed_values())) {
+    return Type::Interface("TNumber", "org.tensorflow.types.family");
+  }
+  return Type::Interface("TType", "org.tensorflow.types.family");
+}
+
 string SnakeToCamelCase(const string& str, bool upper = false) {
   string result;
   bool cap = upper;

tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc

@@ -305,8 +305,15 @@ SourceWriter& SourceWriter::WriteGenerics(
       Append(", ");
     }
     Append(pt->name());
-    if (!pt->supertypes().empty()) {
-      Append(" extends ").AppendType(pt->supertypes().front());
+    bool first_bound = true;
+    for (const Type& bound : pt->supertypes()) {
+      if (first_bound) {
+        Append(" extends ");
+        first_bound = false;
+      } else {
+        Append(" & ");
+      }
+      AppendType(bound);
     }
     first = false;
   }

tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java

@@ -23,7 +23,6 @@
 import org.tensorflow.EagerSession;
 import org.tensorflow.ExecutionEnvironment;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.DoubleNdArray;
@@ -1052,6 +1051,17 @@ public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> bou
     return Bucketize.create(scope, input, boundaries);
   }
 
+  /**
+   * Create a constant from a Tensor.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param tensor a Tensor holding the constant value
+   * @return a constant of the same data type as `tensor`
+   */
+  public <T extends TType> Constant<T> capture(T tensor) {
+    return Constant.create(scope, tensor);
+  }
+
   /**
    * Clips tensor values to a specified min and max.
    *  <p>
@@ -1687,17 +1697,6 @@ public Constant<TInt64> constant(Shape shape) {
     return Constant.tensorOf(scope, shape);
   }
 
-  /**
-   * Create a constant from a Tensor.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param tensor a Tensor holding the constant value
-   * @return a constant of the same data type as `tensor`
-   */
-  public <T extends TType> Constant<T> constant(Tensor<T> tensor) {
-    return Constant.create(scope, tensor);
-  }
-
   /**
    * Creates a constant of {@code String} elements, using the given charset.
    *

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java

@@ -93,7 +93,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AudioSpectrogram create(Scope scope, Operand<TFloat32> input, Long windowSize, Long stride, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AudioSpectrogram", scope.makeOpName("AudioSpectrogram"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("window_size", windowSize);
     opBuilder.setAttr("stride", stride);
@@ -123,7 +123,7 @@ public Output<TFloat32> spectrogram() {
   }
 
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return spectrogram;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java

@@ -89,7 +89,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeWav create(Scope scope, Operand<TString> contents, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeWav", scope.makeOpName("DecodeWav"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java

@@ -54,8 +54,8 @@ public final class EncodeWav extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static EncodeWav create(Scope scope, Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeWav", scope.makeOpName("EncodeWav"));
-    opBuilder.addInput(audio.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(audio.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new EncodeWav(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<TString> contents() {
   }
 
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java

@@ -102,8 +102,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Mfcc create(Scope scope, Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mfcc", scope.makeOpName("Mfcc"));
-    opBuilder.addInput(spectrogram.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(spectrogram.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -161,7 +161,7 @@ public Output<TFloat32> output() {
   }
 
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return output;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java

@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise AND of `x` and `y`.
@@ -67,8 +66,8 @@ public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operan
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseAnd", scope.makeOpName("BitwiseAnd"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseAnd<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java

@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise OR of `x` and `y`.
@@ -67,8 +66,8 @@ public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseOr", scope.makeOpName("BitwiseOr"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseOr<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java

@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise XOR of `x` and `y`.
@@ -67,8 +66,8 @@ public final class BitwiseXor<T extends TNumber> extends RawOp implements Operan
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseXor", scope.makeOpName("BitwiseXor"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseXor<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
 

tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java

@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
@@ -87,7 +86,7 @@ public final class Invert<T extends TNumber> extends RawOp implements Operand<T>
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> Invert<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Invert", scope.makeOpName("Invert"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Invert<T>(opBuilder.build());
   }
@@ -99,7 +98,7 @@ public Output<T> y() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }