mkaravel commented on code in PR #40615: URL: https://github.com/apache/spark/pull/40615#discussion_r1180891177
########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/datasketchesAggregates.scala: ########## @@ -0,0 +1,368 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.SketchesArgumentException +import org.apache.datasketches.hll.{HllSketch, Union} +import org.apache.datasketches.memory.Memory + +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.trees.BinaryLike +import org.apache.spark.sql.types.{AbstractDataType, BinaryType, BooleanType, DataType, IntegerType, LongType, StringType, TypeCollection} +import org.apache.spark.unsafe.types.UTF8String + + +/** + * The HllSketchAgg function utilizes a Datasketches HllSketch instance to + * probabilistically count the number of unique values in a given column, and + * outputs the binary representation of the HllSketch. + * + * See [[https://datasketches.apache.org/docs/HLL/HLL.html]] for more information + * + * @param child child expression against which unique counting will occur + * @param lgConfigK the log-base-2 of K, where K is the number of buckets or slots for the sketch + */ +// scalastyle:off line.size.limit +@ExpressionDescription( + usage = """ + _FUNC_(expr, lgConfigK) - Returns the HllSketch's updateable binary representation. + `lgConfigK` (optional) the log-base-2 of K, with K is the number of buckets or + slots for the HllSketch. """, + examples = """ + Examples: + > SELECT hll_sketch_estimate(_FUNC_(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); + 3 + """, + group = "agg_funcs", + since = "3.5.0") +// scalastyle:on line.size.limit +case class HllSketchAgg( + child: Expression, + lgConfigKExpression: Expression, + mutableAggBufferOffset: Int = 0, + inputAggBufferOffset: Int = 0) + extends TypedImperativeAggregate[HllSketch] with BinaryLike[Expression] with ExpectsInputTypes { + + // Hllsketch config - mark as lazy so that they're not evaluated during tree transformation. + + lazy val lgConfigK: Int = { + val lgConfigK = lgConfigKExpression.eval().asInstanceOf[Int] + // can't use HllUtil.checkLgK so replicate the check + if (lgConfigK < 4 || lgConfigK > 21) { + throw new SketchesArgumentException("Invalid lgConfigK value") + } else { + lgConfigK + } + } + + // Constructors + + def this(child: Expression) = { + this(child, Literal(HllSketch.DEFAULT_LG_K), 0, 0) + } + + def this(child: Expression, lgConfigK: Expression) = { + this(child, lgConfigK, 0, 0) + } + + def this(child: Expression, lgConfigK: Int) = { + this(child, Literal(lgConfigK), 0, 0) + } + + // Copy constructors required by ImperativeAggregate + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): HllSketchAgg = + copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): HllSketchAgg = + copy(inputAggBufferOffset = newInputAggBufferOffset) + + override protected def withNewChildrenInternal(newLeft: Expression, + newRight: Expression): HllSketchAgg = + copy(child = newLeft, lgConfigKExpression = newRight) + + // Overrides for TernaryLike + + override def left: Expression = child + + override def right: Expression = lgConfigKExpression + + // Overrides for TypedImperativeAggregate + + override def prettyName: String = "hll_sketch_agg" + + override def inputTypes: Seq[AbstractDataType] = + Seq(TypeCollection(IntegerType, LongType, StringType, BinaryType), IntegerType) + + override def dataType: DataType = BinaryType + + override def nullable: Boolean = false + + /** + * Instantiate an HllSketch instance using the lgConfigK param. + * + * @return an HllSketch instance + */ + override def createAggregationBuffer(): HllSketch = { + new HllSketch(lgConfigK) + } + + /** + * Evaluate the input row and update the HllSketch instance with the row's value. + * The update function only supports a subset of Spark SQL types, and an + * UnsupportedOperationException will be thrown for unsupported types. + * + * @param sketch The HllSketch instance. + * @param input an input row Review Comment: nit: "an" -> "An" ########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/datasketchesAggregates.scala: ########## @@ -0,0 +1,368 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.SketchesArgumentException +import org.apache.datasketches.hll.{HllSketch, Union} +import org.apache.datasketches.memory.Memory + +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.trees.BinaryLike +import org.apache.spark.sql.types.{AbstractDataType, BinaryType, BooleanType, DataType, IntegerType, LongType, StringType, TypeCollection} +import org.apache.spark.unsafe.types.UTF8String + + +/** + * The HllSketchAgg function utilizes a Datasketches HllSketch instance to + * probabilistically count the number of unique values in a given column, and + * outputs the binary representation of the HllSketch. + * + * See [[https://datasketches.apache.org/docs/HLL/HLL.html]] for more information + * + * @param child child expression against which unique counting will occur + * @param lgConfigK the log-base-2 of K, where K is the number of buckets or slots for the sketch + */ +// scalastyle:off line.size.limit +@ExpressionDescription( + usage = """ + _FUNC_(expr, lgConfigK) - Returns the HllSketch's updateable binary representation. + `lgConfigK` (optional) the log-base-2 of K, with K is the number of buckets or + slots for the HllSketch. """, + examples = """ + Examples: + > SELECT hll_sketch_estimate(_FUNC_(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); + 3 + """, + group = "agg_funcs", + since = "3.5.0") +// scalastyle:on line.size.limit +case class HllSketchAgg( + child: Expression, + lgConfigKExpression: Expression, + mutableAggBufferOffset: Int = 0, + inputAggBufferOffset: Int = 0) + extends TypedImperativeAggregate[HllSketch] with BinaryLike[Expression] with ExpectsInputTypes { + + // Hllsketch config - mark as lazy so that they're not evaluated during tree transformation. + + lazy val lgConfigK: Int = { + val lgConfigK = lgConfigKExpression.eval().asInstanceOf[Int] + // can't use HllUtil.checkLgK so replicate the check + if (lgConfigK < 4 || lgConfigK > 21) { + throw new SketchesArgumentException("Invalid lgConfigK value") + } else { Review Comment: +1 ########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/datasketchesAggregates.scala: ########## @@ -0,0 +1,368 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.SketchesArgumentException +import org.apache.datasketches.hll.{HllSketch, Union} +import org.apache.datasketches.memory.Memory + +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.trees.BinaryLike +import org.apache.spark.sql.types.{AbstractDataType, BinaryType, BooleanType, DataType, IntegerType, LongType, StringType, TypeCollection} +import org.apache.spark.unsafe.types.UTF8String + + +/** + * The HllSketchAgg function utilizes a Datasketches HllSketch instance to + * probabilistically count the number of unique values in a given column, and + * outputs the binary representation of the HllSketch. + * + * See [[https://datasketches.apache.org/docs/HLL/HLL.html]] for more information + * + * @param child child expression against which unique counting will occur + * @param lgConfigK the log-base-2 of K, where K is the number of buckets or slots for the sketch + */ +// scalastyle:off line.size.limit +@ExpressionDescription( + usage = """ + _FUNC_(expr, lgConfigK) - Returns the HllSketch's updateable binary representation. + `lgConfigK` (optional) the log-base-2 of K, with K is the number of buckets or + slots for the HllSketch. """, + examples = """ + Examples: + > SELECT hll_sketch_estimate(_FUNC_(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); + 3 + """, + group = "agg_funcs", + since = "3.5.0") +// scalastyle:on line.size.limit +case class HllSketchAgg( + child: Expression, + lgConfigKExpression: Expression, + mutableAggBufferOffset: Int = 0, + inputAggBufferOffset: Int = 0) + extends TypedImperativeAggregate[HllSketch] with BinaryLike[Expression] with ExpectsInputTypes { + + // Hllsketch config - mark as lazy so that they're not evaluated during tree transformation. + + lazy val lgConfigK: Int = { + val lgConfigK = lgConfigKExpression.eval().asInstanceOf[Int] + // can't use HllUtil.checkLgK so replicate the check + if (lgConfigK < 4 || lgConfigK > 21) { + throw new SketchesArgumentException("Invalid lgConfigK value") + } else { + lgConfigK + } + } + + // Constructors + + def this(child: Expression) = { + this(child, Literal(HllSketch.DEFAULT_LG_K), 0, 0) + } + + def this(child: Expression, lgConfigK: Expression) = { + this(child, lgConfigK, 0, 0) + } + + def this(child: Expression, lgConfigK: Int) = { + this(child, Literal(lgConfigK), 0, 0) + } + + // Copy constructors required by ImperativeAggregate + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): HllSketchAgg = + copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): HllSketchAgg = + copy(inputAggBufferOffset = newInputAggBufferOffset) + + override protected def withNewChildrenInternal(newLeft: Expression, + newRight: Expression): HllSketchAgg = + copy(child = newLeft, lgConfigKExpression = newRight) + + // Overrides for TernaryLike Review Comment: Seems like this is leftover. It should be "BinaryLike" ########## sql/core/src/main/scala/org/apache/spark/sql/functions.scala: ########## @@ -597,6 +597,103 @@ object functions { grouping_id((Seq(colName) ++ colNames).map(n => Column(n)) : _*) } + /** + * Aggregate function: returns the compact binary representation of the Datasketches Review Comment: I believe we are returning the updatable representation now. Same for the functions below. ########## python/pyspark/sql/functions.py: ########## @@ -10113,6 +10113,157 @@ def unwrap_udt(col: "ColumnOrName") -> Column: return _invoke_function("unwrap_udt", _to_java_column(col)) +@try_remote_functions +def hll_sketch_agg(col: "ColumnOrName", lgConfigK: Optional[int] = None) -> Column: + """ + Aggregate function: returns the updatable binary representation of the Datasketches + HllSketch configured with lgConfigK arg. + + .. versionadded:: 3.5.0 + + Parameters + ---------- + col : :class:`~pyspark.sql.Column` or str + lgConfigK : int, optional + The log-base-2 of K, where K is the number of buckets or slots for the HllSketch + + Returns + ------- + :class:`~pyspark.sql.Column` + The binary representation of the HllSketch. + + Examples + -------- + >>> df = spark.createDataFrame([1,2,2,3], "INT") + >>> df = df.agg(hll_sketch_estimate(hll_sketch_agg("value")).alias("distinct_cnt")) + >>> df.show() + +------------+ + |distinct_cnt| + +------------+ + | 3| + +------------+ + """ + if lgConfigK is not None: + return _invoke_function("hll_sketch_agg", _to_java_column(col), lgConfigK) + else: + return _invoke_function("hll_sketch_agg", _to_java_column(col)) + + +@try_remote_functions +def hll_union_agg(col: "ColumnOrName", allowDifferentLgConfigK: Optional[bool] = None) -> Column: + """ + Aggregate function: returns the updaable binary representation of the Datasketches Review Comment: Typo: "updaable" -> "updatable" ########## connector/connect/client/jvm/src/main/scala/org/apache/spark/sql/functions.scala: ########## @@ -539,6 +539,92 @@ object functions { def grouping_id(colName: String, colNames: String*): Column = grouping_id((Seq(colName) ++ colNames).map(n => Column(n)): _*) + /** + * Aggregate function: returns the compact binary representation of the Datasketches HllSketch Review Comment: Here and below: I believe we are now returning the updatable sketch representation. ########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/datasketchesAggregates.scala: ########## @@ -0,0 +1,368 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.SketchesArgumentException +import org.apache.datasketches.hll.{HllSketch, Union} +import org.apache.datasketches.memory.Memory + +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.trees.BinaryLike +import org.apache.spark.sql.types.{AbstractDataType, BinaryType, BooleanType, DataType, IntegerType, LongType, StringType, TypeCollection} +import org.apache.spark.unsafe.types.UTF8String + + +/** + * The HllSketchAgg function utilizes a Datasketches HllSketch instance to + * probabilistically count the number of unique values in a given column, and Review Comment: nit: Forgive me for making this comment (this is my math/algorithms background kicking in): "to probabilistically count": when I read this I under stand that the algorithm for computing the number of unique values is probabilistic (or randomized if you want). I think what we want to convey here is that the count is based on probabilistic analysis. How about: "to count a probabilistic approximation of the number of unique values...". ########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/datasketchesAggregates.scala: ########## @@ -0,0 +1,368 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.SketchesArgumentException +import org.apache.datasketches.hll.{HllSketch, Union} +import org.apache.datasketches.memory.Memory + +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.trees.BinaryLike +import org.apache.spark.sql.types.{AbstractDataType, BinaryType, BooleanType, DataType, IntegerType, LongType, StringType, TypeCollection} +import org.apache.spark.unsafe.types.UTF8String + + +/** + * The HllSketchAgg function utilizes a Datasketches HllSketch instance to + * probabilistically count the number of unique values in a given column, and + * outputs the binary representation of the HllSketch. + * + * See [[https://datasketches.apache.org/docs/HLL/HLL.html]] for more information + * + * @param child child expression against which unique counting will occur + * @param lgConfigK the log-base-2 of K, where K is the number of buckets or slots for the sketch + */ +// scalastyle:off line.size.limit +@ExpressionDescription( + usage = """ + _FUNC_(expr, lgConfigK) - Returns the HllSketch's updateable binary representation. + `lgConfigK` (optional) the log-base-2 of K, with K is the number of buckets or + slots for the HllSketch. """, + examples = """ + Examples: + > SELECT hll_sketch_estimate(_FUNC_(col, 12)) FROM VALUES (1), (1), (2), (2), (3) tab(col); + 3 + """, + group = "agg_funcs", + since = "3.5.0") +// scalastyle:on line.size.limit +case class HllSketchAgg( + child: Expression, + lgConfigKExpression: Expression, + mutableAggBufferOffset: Int = 0, + inputAggBufferOffset: Int = 0) + extends TypedImperativeAggregate[HllSketch] with BinaryLike[Expression] with ExpectsInputTypes { + + // Hllsketch config - mark as lazy so that they're not evaluated during tree transformation. + + lazy val lgConfigK: Int = { + val lgConfigK = lgConfigKExpression.eval().asInstanceOf[Int] + // can't use HllUtil.checkLgK so replicate the check + if (lgConfigK < 4 || lgConfigK > 21) { Review Comment: +1 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
