Github user ron8hu commented on a diff in the pull request:
https://github.com/apache/spark/pull/16395#discussion_r95096610
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/estimation/FilterEstimation.scala
---
@@ -0,0 +1,479 @@
+/*
+ * 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.plans.logical.estimation
+
+import scala.collection.immutable.{HashSet, Map}
+import scala.collection.mutable
+
+import org.apache.spark.internal.Logging
+import org.apache.spark.sql.catalyst.expressions._
+import org.apache.spark.sql.catalyst.plans.logical._
+import org.apache.spark.sql.catalyst.util.DateTimeUtils
+import org.apache.spark.sql.types._
+import org.apache.spark.unsafe.types.UTF8String
+
+
+object FilterEstimation extends Logging {
+
+ /**
+ * We use a mutable colStats because we need to update the corresponding
ColumnStat
+ * for a column after we apply a predicate condition.
+ */
+ private var mutableColStats: mutable.Map[ExprId, ColumnStat] =
mutable.Map.empty
+
+ def estimate(plan: Filter): Option[Statistics] = {
+ val stats: Statistics = plan.child.statistics
+ if (stats.rowCount.isEmpty) return None
+
+ /** save a mutable copy of colStats so that we can later change it
recursively */
+ val statsExprIdMap: Map[ExprId, ColumnStat] =
+ stats.attributeStats.map(kv => (kv._1.exprId, kv._2))
+ mutableColStats = mutable.Map.empty ++= statsExprIdMap
+
+ /** save a copy of ExprId-to-Attribute map for later conversion use */
+ val expridToAttrMap: Map[ExprId, Attribute] =
+ stats.attributeStats.map(kv => (kv._1.exprId, kv._1))
+
+ /** estimate selectivity for this filter */
+ val percent: Double = calculateConditions(plan, plan.condition)
+
+ /** copy mutableColStats contents to an immutable AttributeMap */
+ val mutableAttributeStats: mutable.Map[Attribute, ColumnStat] =
+ mutableColStats.map(kv => (expridToAttrMap(kv._1) -> kv._2))
+ val newColStats = AttributeMap(mutableAttributeStats.toSeq)
+
+ val filteredRowCountValue: BigInt =
+ EstimationUtils.ceil(BigDecimal(stats.rowCount.get) * percent)
+ val avgRowSize = BigDecimal(EstimationUtils.getRowSize(plan.output,
newColStats))
+ val filteredSizeInBytes: BigInt =
+ EstimationUtils.ceil(BigDecimal(filteredRowCountValue) * avgRowSize)
+
+ Some(stats.copy(sizeInBytes = filteredSizeInBytes, rowCount =
Some(filteredRowCountValue),
+ attributeStats = newColStats))
+ }
+
+ def calculateConditions(
+ plan: Filter,
+ condition: Expression,
+ update: Boolean = true)
+ : Double = {
+ /**
+ * For conditions linked by And, we need to update stats after a
condition estimation
+ * so that the stats will be more accurate for subsequent estimation.
+ * For conditions linked by OR, we do not update stats after a
condition estimation.
+ */
+ condition match {
+ case And(cond1, cond2) =>
+ val p1 = calculateConditions(plan, cond1, update)
+ val p2 = calculateConditions(plan, cond2, update)
+ p1 * p2
+
+ case Or(cond1, cond2) =>
+ val p1 = calculateConditions(plan, cond1, update = false)
+ val p2 = calculateConditions(plan, cond2, update = false)
+ math.min(1.0, p1 + p2 - (p1 * p2))
+
+ case Not(cond) => calculateSingleCondition(plan, cond, isNot = true,
update = false)
+ case _ => calculateSingleCondition(plan, condition, isNot = false,
update)
+ }
+ }
+
+ def calculateSingleCondition(
+ plan: Filter,
+ condition: Expression,
+ isNot: Boolean,
+ update: Boolean)
+ : Double = {
+ var notSupported: Boolean = false
+ val percent: Double = condition match {
+ /**
+ * Currently we only support binary predicates where one side is a
column,
+ * and the other is a literal.
+ * Note that: all binary predicate computing methods assume the
literal is at the right side,
+ * so we will change the predicate order if not.
+ */
+ case op@LessThan(ExtractAttr(ar), l: Literal) =>
+ evaluateBinary(op, ar, l, update)
+ case op@LessThan(l: Literal, ExtractAttr(ar)) =>
+ evaluateBinary(GreaterThan(ar, l), ar, l, update)
+
+ case op@LessThanOrEqual(ExtractAttr(ar), l: Literal) =>
+ evaluateBinary(op, ar, l, update)
+ case op@LessThanOrEqual(l: Literal, ExtractAttr(ar)) =>
+ evaluateBinary(GreaterThanOrEqual(ar, l), ar, l, update)
+
+ case op@GreaterThan(ExtractAttr(ar), l: Literal) =>
+ evaluateBinary(op, ar, l, update)
+ case op@GreaterThan(l: Literal, ExtractAttr(ar)) =>
+ evaluateBinary(LessThan(ar, l), ar, l, update)
+
+ case op@GreaterThanOrEqual(ExtractAttr(ar), l: Literal) =>
+ evaluateBinary(op, ar, l, update)
+ case op@GreaterThanOrEqual(l: Literal, ExtractAttr(ar)) =>
+ evaluateBinary(LessThanOrEqual(ar, l), ar, l, update)
+
+ /** EqualTo does not care about the order */
+ case op@EqualTo(ExtractAttr(ar), l: Literal) =>
+ evaluateBinary(op, ar, l, update)
+ case op@EqualTo(l: Literal, ExtractAttr(ar)) =>
+ evaluateBinary(op, ar, l, update)
+
+ case In(ExtractAttr(ar), expList) if
!expList.exists(!_.isInstanceOf[Literal]) =>
+ /**
+ * Expression [In (value, seq[Literal])] will be replaced with
optimized version
+ * [InSet (value, HashSet[Literal])] in Optimizer, but only for
list.size > 10.
+ * Here we convert In into InSet anyway, because they share the
same processing logic.
+ */
+ val hSet = expList.map(e => e.eval())
+ evaluateInSet(ar, HashSet() ++ hSet, update)
+
+ case InSet(ExtractAttr(ar), set) =>
+ evaluateInSet(ar, set, update)
+
+ /**
+ * It's difficult to estimate IsNull after outer joins. Hence,
+ * we support IsNull and IsNotNull only when the child is a leaf
node (table).
+ */
+ case IsNull(ExtractAttr(ar)) =>
+ if (plan.child.isInstanceOf[LeafNode ]) {
+ evaluateIsNull(plan, ar, true, update)
+ }
+ else 1.0
+
+ case IsNotNull(ExtractAttr(ar)) =>
+ if (plan.child.isInstanceOf[LeafNode ]) {
+ evaluateIsNull(plan, ar, false, update)
+ }
+ else 1.0
+
+ case _ =>
+ /**
+ * TODO: it's difficult to support string operators without
advanced statistics.
+ * Hence, these string operators Like(_, _) | Contains(_, _) |
StartsWith(_, _)
+ * | EndsWith(_, _) are not supported yet
+ */
+ logDebug("[CBO] Unsupported filter condition: " + condition)
+ notSupported = true
+ 1.0
+ }
+ if (notSupported) {
+ 1.0
+ } else if (isNot) {
+ 1.0 - percent
+ } else {
+ percent
+ }
+ }
+
+ def evaluateIsNull(
+ plan: Filter,
+ attrRef: AttributeReference,
+ isNull: Boolean,
+ update: Boolean)
+ : Double = {
+ if (!mutableColStats.contains(attrRef.exprId)) {
+ logDebug("[CBO] No statistics for " + attrRef)
+ return 1.0
+ }
+ val aColStat = mutableColStats(attrRef.exprId)
+ val rowCountValue = plan.child.statistics.rowCount.get
+ val nullPercent: BigDecimal =
+ if (rowCountValue == 0) 0.0
+ else BigDecimal(aColStat.nullCount)/BigDecimal(rowCountValue)
+
+ if (update) {
+ val newStats =
+ if (isNull) aColStat.copy(distinctCount = 0, min = None, max =
None)
+ else aColStat.copy(nullCount = 0)
+
+ mutableColStats += (attrRef.exprId -> newStats)
+ }
+
+ val percent =
+ if (isNull) nullPercent.toDouble
+ else {
+ /** ISNOTNULL(column) */
+ 1.0 - nullPercent.toDouble
+ }
+
+ percent
+ }
+
+ /** This method evaluates binary comparison operators such as =, <, <=,
>, >= */
+ def evaluateBinary(
+ op: BinaryComparison,
+ attrRef: AttributeReference,
+ literal: Literal,
+ update: Boolean)
+ : Double = {
+ if (!mutableColStats.contains(attrRef.exprId)) {
+ logDebug("[CBO] No statistics for " + attrRef)
+ return 1.0
+ }
+
+ /** Make sure that the Date/Timestamp literal is a valid one */
+ attrRef.dataType match {
+ case DateType =>
+ val dateLiteral =
DateTimeUtils.stringToDate(literal.value.asInstanceOf[UTF8String])
+ if (dateLiteral.isEmpty) {
+ logDebug("[CBO] Date literal is wrong, No statistics for " +
attrRef)
+ return 1.0
+ }
+ case TimestampType =>
+ val tsLiteral =
DateTimeUtils.stringToTimestamp(literal.value.asInstanceOf[UTF8String])
+ if (tsLiteral.isEmpty) {
+ logDebug("[CBO] Timestamp literal is wrong, No statistics for "
+ attrRef)
+ return 1.0
+ }
+ case _ =>
+ }
+
+ op match {
+ case EqualTo(l, r) => evaluateEqualTo(op, attrRef, literal, update)
+ case _ =>
+ attrRef.dataType match {
+ case _: NumericType | DateType | TimestampType =>
+ evaluateBinaryForNumeric(op, attrRef, literal, update)
+ case StringType | BinaryType =>
+ /**
+ * TODO: It is difficult to support other binary comparisons
for String/Binary
+ * type without min/max and advanced statistics like histogram.
+ */
+ logDebug("[CBO] No statistics for String/Binary type " +
attrRef)
+ return 1.0
+ }
+ }
+ }
+
+ /**
+ * This method converts a numeric or Literal value of numeric type to a
BigDecimal value.
+ * If isNumeric is true, then it is a numeric value. Otherwise, it is a
Literal value.
+ */
+ def numericLiteralToBigDecimal(
+ literal: Any,
+ dataType: DataType,
+ isNumeric: Boolean = false)
+ : BigDecimal = {
+ dataType match {
+ case _: IntegralType =>
+ if (isNumeric) BigDecimal(literal.asInstanceOf[Long])
+ else
BigDecimal(literal.asInstanceOf[Literal].value.asInstanceOf[Long])
+ case _: FractionalType =>
+ if (isNumeric) BigDecimal(literal.asInstanceOf[Double])
+ else
BigDecimal(literal.asInstanceOf[Literal].value.asInstanceOf[Double])
+ case DateType =>
+ if (isNumeric) BigDecimal(literal.asInstanceOf[BigInt])
+ else {
+ val dateLiteral = DateTimeUtils.stringToDate(
+ literal.asInstanceOf[Literal].value.asInstanceOf[UTF8String])
+ BigDecimal(dateLiteral.asInstanceOf[BigInt])
+ }
+ case TimestampType =>
+ if (isNumeric) BigDecimal(literal.asInstanceOf[BigInt])
+ else {
+ val tsLiteral = DateTimeUtils.stringToTimestamp(
+ literal.asInstanceOf[Literal].value.asInstanceOf[UTF8String])
+ BigDecimal(tsLiteral.asInstanceOf[BigInt])
+ }
+ }
+ }
+
+ /** This method evaluates the equality predicate for all data types. */
+ def evaluateEqualTo(
+ op: BinaryComparison,
+ attrRef: AttributeReference,
+ literal: Literal,
+ update: Boolean)
+ : Double = {
+
+ val aColStat = mutableColStats(attrRef.exprId)
+ val ndv = aColStat.distinctCount
+
+ /**
+ * decide if the value is in [min, max] of the column.
+ * We currently don't store min/max for binary/string type.
+ * Hence, we assume it is in boundary for binary/string type.
+ */
+ val inBoundary: Boolean = attrRef.dataType match {
+ case _: NumericType | DateType | TimestampType =>
+ val statsRange =
+ Range(aColStat.min, aColStat.max,
attrRef.dataType).asInstanceOf[NumericRange]
+ val lit = numericLiteralToBigDecimal(literal, attrRef.dataType)
+ (lit >= statsRange.min) && (lit <= statsRange.max)
+
+ case _ => true /** for String/Binary type */
+ }
+
+ val percent: Double =
+ if (inBoundary) {
+
+ if (update) {
+ /**
+ * We update ColumnStat structure after apply this equality
predicate.
+ * Set distinctCount to 1. Set nullCount to 0.
+ */
+ val newStats = attrRef.dataType match {
+ case _: NumericType | DateType | TimestampType =>
+ val newValue = Some(literal.value)
+ aColStat.copy(distinctCount = 1, min = newValue,
+ max = newValue, nullCount = 0)
+ case _ => aColStat.copy(distinctCount = 1, nullCount = 0)
+ }
+ mutableColStats += (attrRef.exprId -> newStats)
+ }
+
+ 1.0 / ndv.toDouble
+ } else {
+ 0.0
+ }
+
+ percent
+ }
+
+ def evaluateInSet(
+ attrRef: AttributeReference,
+ hSet: Set[Any],
+ update: Boolean)
+ : Double = {
+ if (!mutableColStats.contains(attrRef.exprId)) {
+ logDebug("[CBO] No statistics for " + attrRef)
+ return 1.0
+ }
+
+ val aColStat = mutableColStats(attrRef.exprId)
+ val ndv = aColStat.distinctCount
+ val aType = attrRef.dataType
+
+ // use [min, max] to filter the original hSet
+ val validQuerySet = aType match {
+ case _: NumericType | DateType | TimestampType =>
+ val statsRange =
+ Range(aColStat.min, aColStat.max,
aType).asInstanceOf[NumericRange]
+ hSet.map(e => numericLiteralToBigDecimal(e, aType, true)).
+ filter(e => e >= statsRange.min && e <= statsRange.max)
+
+ /** We assume the whole set since there is no min/max information
for String/Binary type */
+ case StringType | BinaryType => hSet
+ }
+ if (validQuerySet.isEmpty) {
+ return 0.0
+ }
+
+ val newNdv = validQuerySet.size
+ val(newMax, newMin) = aType match {
+ case _: NumericType | DateType | TimestampType =>
+ val tmpSet: Set[Double] = validQuerySet.map(e =>
e.toString.toDouble)
+ (Some(tmpSet.max), Some(tmpSet.min))
+ case _ =>
+ (None, None)
+ }
+
+ if (update) {
+ val newStats = attrRef.dataType match {
+ case _: NumericType | DateType | TimestampType =>
+ aColStat.copy(distinctCount = newNdv, min = newMin,
+ max = newMax, nullCount = 0)
+ case StringType | BinaryType =>
+ aColStat.copy(distinctCount = newNdv, nullCount = 0)
+ }
+ mutableColStats += (attrRef.exprId -> newStats)
+ }
+
+ /**
+ * return the filter selectivity. Without advanced statistics such as
histograms,
+ * we have to assume uniform distribution.
+ */
+ math.min(1.0, validQuerySet.size / ndv.toDouble)
+ }
+
+ def evaluateBinaryForNumeric(
--- End diff --
Yes, the return value is a double value showing the percentage of rows
meeting a given condition. Also I will add comments for this method in JavaDoc
style.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
