385 chained joins - possible solution

dataset/src/main/scala/frameless/FramelessSyntax.scala

@@ -1,5 +1,6 @@
 package frameless
 
+import frameless.ops.ChainedJoinOps
 import org.apache.spark.sql.{Column, DataFrame, Dataset}
 
 trait FramelessSyntax {
@@ -15,4 +16,8 @@ trait FramelessSyntax {
   implicit class DataframeSyntax(self: DataFrame){
     def unsafeTyped[T: TypedEncoder]: TypedDataset[T] = TypedDataset.createUnsafe(self)
   }
+
+  implicit class ChainedJoinSyntax[T](ds: TypedDataset[T]) {
+    def join[U](other: TypedDataset[U]): ChainedJoinOps[T, U] = new ChainedJoinOps[T, U](ds, other)
+  }
 }

dataset/src/main/scala/frameless/ops/ChainedJoinOps.scala

@@ -0,0 +1,159 @@
+package frameless.ops
+
+import frameless.{TypedColumn, TypedDataset, TypedEncoder}
+
+/**
+ * Collection of forwarding functions that optionally provide a reference to the incoming dataset for chaining of joins
+ * @param ds the dataset on which .join(other) was called
+ * @param other the dataset to which ds is joined
+ * @tparam T the type of ds
+ * @tparam U the type of other
+ */
+case class ChainedJoinOps[T, U](ds: TypedDataset[T], other: TypedDataset[U]) {
+  /** Computes the right outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def right(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean])(implicit e: TypedEncoder[(Option[T], U)]): TypedDataset[(Option[T], U)] =
+    ds.joinRight(other)(conditionF(ds))
+
+  /** Computes the right outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def right(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean])(implicit e: TypedEncoder[(Option[T], U)]): TypedDataset[(Option[T], U)] =
+    ds.joinRight(other)(conditionF(ds, other))
+
+  /** Computes the right outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def right(condition: TypedColumn[T with U, Boolean])(implicit e: TypedEncoder[(Option[T], U)]): TypedDataset[(Option[T], U)] =
+    ds.joinRight(other)(condition)
+
+  /** Computes the cartesian project of `this` `Dataset` with the `other` `Dataset` */
+  def cross() // here for completeness
+                  (implicit e: TypedEncoder[(T, U)]): TypedDataset[(T, U)] =
+    ds.joinCross(other)
+
+  /** Computes the full outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def full(condition: TypedColumn[T with U, Boolean])
+                 (implicit e: TypedEncoder[(Option[T], Option[U])]): TypedDataset[(Option[T], Option[U])] =
+    ds.joinFull(other)(condition)
+
+  /** Computes the full outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def full(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean])
+             (implicit e: TypedEncoder[(Option[T], Option[U])]): TypedDataset[(Option[T], Option[U])] =
+    ds.joinFull(other)(conditionF(ds))
+
+  /** Computes the full outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def full(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean])
+          (implicit e: TypedEncoder[(Option[T], Option[U])]): TypedDataset[(Option[T], Option[U])] =
+    ds.joinFull(other)(conditionF(ds, other))
+
+  /** Computes the inner join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def inner(condition: TypedColumn[T with U, Boolean])
+                  (implicit e: TypedEncoder[(T, U)]): TypedDataset[(T, U)] =
+    ds.joinInner(other)(condition)
+
+  /** Computes the inner join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def inner(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean])
+              (implicit e: TypedEncoder[(T, U)]): TypedDataset[(T, U)] =
+    ds.joinInner(other)(conditionF(ds))
+
+  /** Computes the inner join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def inner(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean])
+           (implicit e: TypedEncoder[(T, U)]): TypedDataset[(T, U)] =
+    ds.joinInner(other)(conditionF(ds, other))
+
+  /** Computes the left outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def left(condition: TypedColumn[T with U, Boolean])
+                 (implicit e: TypedEncoder[(T, Option[U])]): TypedDataset[(T, Option[U])] =
+    ds.joinLeft(other)(condition)
+
+  /** Computes the left outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def left(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean])
+          (implicit e: TypedEncoder[(T, Option[U])]): TypedDataset[(T, Option[U])] =
+    ds.joinLeft(other)(conditionF(ds))
+
+  /** Computes the left outer join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def left(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean])
+          (implicit e: TypedEncoder[(T, Option[U])]): TypedDataset[(T, Option[U])] =
+    ds.joinLeft(other)(conditionF(ds,other))
+
+  /** Computes the left semi join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def leftSemi(condition: TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftSemi(other)(condition)
+
+  /** Computes the left semi join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def leftSemi(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftSemi(other)(conditionF(ds))
+
+  /** Computes the left semi join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def leftSemi(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftSemi(other)(conditionF(ds, other))
+
+  /** Computes the left anti join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   */
+  def leftAnti(condition: TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftAnti(other)(condition)
+
+  /** Computes the left anti join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current dataset in the chain to the conditionF allowing you access to this TypedDataset's columns
+   */
+  def leftAnti(conditionF: TypedDataset[T] => TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftAnti(other)(conditionF(ds))
+
+  /** Computes the left anti join of `this` `Dataset` with the `other` `Dataset`,
+   * returning a `Tuple2` for each pair where condition evaluates to true.
+   *
+   * This version passes in the current and joined datasets in the chain to the conditionF allowing you access to this TypedDataset's columns and the joins
+   */
+  def leftAnti(conditionF: (TypedDataset[T], TypedDataset[U]) => TypedColumn[T with U, Boolean]): TypedDataset[T] =
+    ds.joinLeftAnti(other)(conditionF(ds, other))
+
+}