Laiacano/java extract subsettings

java/src/main/java/com/spotify/featran/java/JFeatureSpec.java

@@ -185,6 +185,61 @@ public JRecordExtractor<T, double[]> extractWithSettingsDouble(String settings)
     return new JRecordExtractor<>(JavaOps.extractWithSettingsSparseLabeledPoint(self, settings));
   }
 
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag)}.
+   */
+  public JRecordExtractor<T, float[]> extractWithSubsetSettingsFloat(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsFloat(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag)}.
+   */
+  public JRecordExtractor<T, double[]> extractWithSubsetSettingsDouble(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsDouble(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag)}.
+   */
+  public JRecordExtractor<T, FloatSparseArray>
+  extractWithSubsetSettingsFloatSparseArray(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsFloatSparseArray(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag).
+   */
+  public JRecordExtractor<T, DoubleSparseArray>
+  extractWithSubsetSettingsDoubleSparseArray(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsDoubleSparseArray(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag)}.
+   */
+  public JRecordExtractor<T, FloatNamedSparseArray>
+  extractWithSubsetSettingsFloatNamedSparseArray(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsFloatNamedSparseArray(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag).
+   */
+  public JRecordExtractor<T, DoubleNamedSparseArray>
+  extractWithSubsetSettingsDoubleNamedSparseArray(String settings) {
+    return new JRecordExtractor<>(
+        JavaOps.extractWithSubsetSettingsDoubleNamedSparseArray(self, settings));
+  }
+
+  /**
+   * Java wrapper for {@link FeatureSpec#extractWithSubsetSettings(String, FeatureBuilder, ClassTag).
+   */
+  public JRecordExtractor<T, Example>
+  extractWithSubsetSettingsExample(String settings) {
+    return new JRecordExtractor<>(JavaOps.extractWithSubsetSettingsExample(self, settings));
+  }
+
   /**
    * Returns internal {@link FeatureSpec}.
    */

java/src/main/scala/com/spotify/featran/java/JavaOps.scala

@@ -58,6 +58,38 @@ private object JavaOps {
     override def pure[A, B: ClassTag](ma: JList[A])(b: B): JList[B] = Collections.singletonList(b)
   }
 
+  def extractWithSubsetSettingsFloat[T](fs: FeatureSpec[T],
+                                        settings: String): RecordExtractor[T, Array[Float]] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsDouble[T](fs: FeatureSpec[T],
+                                         settings: String): RecordExtractor[T, Array[Double]] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsFloatSparseArray[T](
+      fs: FeatureSpec[T],
+      settings: String): RecordExtractor[T, FloatSparseArray] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsDoubleSparseArray[T](
+      fs: FeatureSpec[T],
+      settings: String): RecordExtractor[T, DoubleSparseArray] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsDoubleNamedSparseArray[T](
+      fs: FeatureSpec[T],
+      settings: String): RecordExtractor[T, DoubleNamedSparseArray] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsFloatNamedSparseArray[T](
+      fs: FeatureSpec[T],
+      settings: String): RecordExtractor[T, FloatNamedSparseArray] =
+    fs.extractWithSubsetSettings(settings)
+
+  def extractWithSubsetSettingsExample[T](fs: FeatureSpec[T],
+                                          settings: String): RecordExtractor[T, Example] =
+    fs.extractWithSubsetSettings(settings)
+
   //================================================================================
   // Wrappers for FeatureSpec
   //================================================================================

java/src/test/scala/com/spotify/featran/java/JavaTest.scala

@@ -144,6 +144,43 @@ class JavaTest extends FlatSpec with Matchers {
     TestData.map(f9.featureValue).map(l2v(ExpectedNames.size)) shouldBe ExpectedValues
   }
 
+  it should "work with extractWithSubsetSettings and RecordExtractor" in {
+    val ExpectedNamesSubset = Seq("one_hot_a")
+    val ExpectedValuesSubset = Seq(1.0, 0.0, 0.0, 0.0, 0.0, 0.00)
+
+    val TestSpecSubset = TestSpec.filter(feature => feature.transformer.name == "one_hot_a")
+
+    val fs = JFeatureSpec.wrap(TestSpecSubset)
+    val settings = fs.extract(TestData.asJava).featureSettings()
+    val f1 = fs.extractWithSubsetSettingsFloat(settings)
+    val f2 = fs.extractWithSubsetSettingsDouble(settings)
+    val f3 = fs.extractWithSubsetSettingsFloatSparseArray(settings)
+    val f4 = fs.extractWithSubsetSettingsDoubleSparseArray(settings)
+    val f5 = fs.extractWithSubsetSettingsFloatNamedSparseArray(settings)
+    val f6 = fs.extractWithSubsetSettingsDoubleNamedSparseArray(settings)
+    val f7 = fs.extractWithSubsetSettingsExample(settings)
+    val f8 = fs.extractWithSubsetSettingsLabeledPoint(settings)
+    val f9 = fs.extractWithSubsetSettingsSparseLabeledPoint(settings)
+    f1.featureNames().asScala shouldBe ExpectedNamesSubset
+    f2.featureNames().asScala shouldBe ExpectedNamesSubset
+    f3.featureNames().asScala shouldBe ExpectedNamesSubset
+    f4.featureNames().asScala shouldBe ExpectedNamesSubset
+    f5.featureNames().asScala shouldBe ExpectedNamesSubset
+    f6.featureNames().asScala shouldBe ExpectedNamesSubset
+    f7.featureNames().asScala shouldBe ExpectedNamesSubset
+    f8.featureNames().asScala shouldBe ExpectedNamesSubset
+    f9.featureNames().asScala shouldBe ExpectedNamesSubset
+    TestData.map(f1.featureValue).map(_.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f2.featureValue).map(_.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f3.featureValue).map(_.toDense.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f4.featureValue).map(_.toDense.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f5.featureValue).map(_.toDense.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f6.featureValue).map(_.toDense.toSeq) shouldBe ExpectedValuesSubset
+    TestData.map(f7.featureValue).map(e2v(ExpectedNamesSubset)) shouldBe ExpectedValuesSubset
+    TestData.map(f8.featureValue).map(_.values.toSeq.map(_.toDouble)) shouldBe ExpectedValuesSubset
+    TestData.map(f9.featureValue).map(l2v(ExpectedNamesSubset.size)) shouldBe ExpectedValuesSubset
+  }
+
   it should "work with sparse arrays" in {
     val in = Seq("a", "b")
     val indices = Seq(Seq(0), Seq(1))