ajantha-bhat commented on a change in pull request #3375: URL: https://github.com/apache/iceberg/pull/3375#discussion_r741182113
########## File path: spark/v3.2/spark/src/main/java/org/apache/iceberg/spark/SparkExpressions.java ########## @@ -0,0 +1,322 @@ +/* + * 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.iceberg.spark; + +import java.time.Instant; +import java.time.LocalDate; +import java.time.LocalDateTime; +import java.time.OffsetDateTime; +import java.time.ZoneOffset; +import java.time.temporal.ChronoUnit; +import java.util.Collection; +import java.util.List; +import java.util.Map; +import java.util.Set; +import java.util.function.Function; +import org.apache.iceberg.expressions.Expression.Operation; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableMap; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.types.Types.TimestampType; +import org.apache.iceberg.util.Pair; +import org.apache.spark.sql.catalyst.analysis.UnresolvedAttribute; +import org.apache.spark.sql.catalyst.expressions.And; +import org.apache.spark.sql.catalyst.expressions.Attribute; +import org.apache.spark.sql.catalyst.expressions.AttributeReference; +import org.apache.spark.sql.catalyst.expressions.BinaryExpression; +import org.apache.spark.sql.catalyst.expressions.Cast; +import org.apache.spark.sql.catalyst.expressions.EqualNullSafe; +import org.apache.spark.sql.catalyst.expressions.EqualTo; +import org.apache.spark.sql.catalyst.expressions.Expression; +import org.apache.spark.sql.catalyst.expressions.GreaterThan; +import org.apache.spark.sql.catalyst.expressions.GreaterThanOrEqual; +import org.apache.spark.sql.catalyst.expressions.In; +import org.apache.spark.sql.catalyst.expressions.InSet; +import org.apache.spark.sql.catalyst.expressions.IsNotNull; +import org.apache.spark.sql.catalyst.expressions.IsNull; +import org.apache.spark.sql.catalyst.expressions.LessThan; +import org.apache.spark.sql.catalyst.expressions.LessThanOrEqual; +import org.apache.spark.sql.catalyst.expressions.Literal; +import org.apache.spark.sql.catalyst.expressions.Not; +import org.apache.spark.sql.catalyst.expressions.Or; +import org.apache.spark.sql.catalyst.expressions.ParseToDate; +import org.apache.spark.sql.catalyst.expressions.UnaryExpression; +import org.apache.spark.sql.catalyst.expressions.Year; +import org.apache.spark.sql.types.DateType$; +import org.apache.spark.sql.types.Decimal; +import org.apache.spark.unsafe.types.UTF8String; +import scala.collection.JavaConverters; + +import static org.apache.iceberg.expressions.Expressions.alwaysFalse; +import static org.apache.iceberg.expressions.Expressions.and; +import static org.apache.iceberg.expressions.Expressions.equal; +import static org.apache.iceberg.expressions.Expressions.not; +import static org.apache.iceberg.expressions.Expressions.or; +import static org.apache.iceberg.expressions.Expressions.predicate; + + +/** + * To convert spark catalyst Expression to Iceberg expression + * + */ +public class SparkExpressions { + private static final Map<Class<? extends Expression>, Operation> FILTERS = ImmutableMap + .<Class<? extends Expression>, Operation>builder() + .put(EqualTo.class, Operation.EQ) + .put(EqualNullSafe.class, Operation.EQ) + .put(GreaterThan.class, Operation.GT) + .put(GreaterThanOrEqual.class, Operation.GT_EQ) + .put(LessThan.class, Operation.LT) + .put(LessThanOrEqual.class, Operation.LT_EQ) + .put(In.class, Operation.IN) + .put(InSet.class, Operation.IN) + .put(IsNull.class, Operation.IS_NULL) + .put(IsNotNull.class, Operation.NOT_NULL) + .put(And.class, Operation.AND) + .put(Or.class, Operation.OR) + .put(Not.class, Operation.NOT) + .build(); + private static final Map<Class<? extends Expression>, Transform> TRANSFORMS = ImmutableMap + .<Class<? extends Expression>, Transform>builder() + .put(UnresolvedAttribute.class, Transform.IDENTITY) + .put(AttributeReference.class, Transform.IDENTITY) + .put(Year.class, Transform.YEAR) + .put(ParseToDate.class, Transform.DAY) + .put(Cast.class, Transform.DAY) + .build(); + private static final OffsetDateTime EPOCH = Instant.ofEpochSecond(0).atOffset(ZoneOffset.UTC); + private static final LocalDate EPOCH_DAY = EPOCH.toLocalDate(); + + private SparkExpressions() { + } + + public static org.apache.iceberg.expressions.Expression convert(Expression expr) { + Class<? extends Expression> exprClass = expr.getClass(); + Operation op = FILTERS.get(exprClass); + if (op != null) { + switch (op) { + case IS_NULL: + case NOT_NULL: + UnaryExpression unary = (UnaryExpression) expr; + if (unary.child() instanceof Attribute) { + Attribute attr = (Attribute) unary.child(); + return predicate(op, attr.name()); + } + return null; + case LT: + case LT_EQ: + case GT: + case GT_EQ: + case EQ: + case NOT_EQ: + BinaryExpression binary = (BinaryExpression) expr; + return convert(op, binary.left(), binary.right()); + case NOT: + org.apache.iceberg.expressions.Expression child = convert(((Not) expr).child()); + if (child != null) { + return not(child); + } + return null; + case AND: + And andExpr = (And) expr; + org.apache.iceberg.expressions.Expression andLeft = convert(andExpr.left()); + org.apache.iceberg.expressions.Expression andRight = convert(andExpr.right()); + if (andLeft != null && andRight != null) { + return and(convert(andExpr.left()), convert(andExpr.right())); + } + return null; + case OR: + Or orExpr = (Or) expr; + org.apache.iceberg.expressions.Expression orLeft = convert(orExpr.left()); + org.apache.iceberg.expressions.Expression orRight = convert(orExpr.right()); + if (orLeft != null && orRight != null) { + return or(orLeft, orRight); + } + return null; + case IN: + if (expr instanceof In) { + In inExpr = (In) expr; + List<Object> literals = convertLiterals(JavaConverters.seqAsJavaListConverter(inExpr.list()).asJava()); + if (literals != null) { + return convertIn(inExpr.value(), literals); + } else { + // if the list contained a non-literal, it can't be converted + return null; + } + } else if (expr instanceof InSet) { + InSet inExpr = (InSet) expr; + // expressions are already converted to Java objects + Set<Object> literals = JavaConverters.setAsJavaSetConverter(inExpr.hset()).asJava(); + return convertIn(inExpr.child(), literals); + } + return null; + default: + } + } + // can't convert + return null; + } + + private static org.apache.iceberg.expressions.Expression convertIn(Expression expr, + Collection<Object> values) { + if (expr instanceof Attribute) { + Attribute attr = (Attribute) expr; + org.apache.iceberg.expressions.Expression converted = alwaysFalse(); + for (Object item : values) { + converted = or(converted, equal(attr.name(), item)); + } + return converted; + } + return null; + } + + private static List<Object> convertLiterals(List<Expression> values) { + List<Object> converted = Lists.newArrayListWithExpectedSize(values.size()); + + for (Expression value : values) { + if (value instanceof Literal) { + Literal lit = (Literal) value; + converted.add(valueFromSpark(lit)); + } else { + return null; + } + } + return converted; + } + + private static org.apache.iceberg.expressions.Expression convert(Operation op, + Expression left, + Expression right) { + Pair<Transform, String> attrPair = null; + Operation leftOperation = null; + Literal lit = null; + + if (right instanceof Literal) { + lit = (Literal) right; + attrPair = convertAttr(left); + leftOperation = op; + } else if (left instanceof Literal) { + lit = (Literal) left; + attrPair = convertAttr(right); + leftOperation = op.flipLR(); + } + + if (attrPair != null) { + switch (attrPair.first()) { + case IDENTITY: + return predicate(leftOperation, attrPair.second(), valueFromSpark(lit)); + case YEAR: + return filter(leftOperation, attrPair.second(), (int) lit.value(), + SparkExpressions::yearToTimestampMicros); + case DAY: + return filter(leftOperation, attrPair.second(), (int) lit.value(), + SparkExpressions::dayToTimestampMicros); + default: + } + } + return null; + } + + private static Object valueFromSpark(Literal lit) { + if (lit.value() instanceof UTF8String) { + return lit.value().toString(); + } else if (lit.value() instanceof Decimal) { + return ((Decimal) lit.value()).toJavaBigDecimal(); + } + return lit.value(); + } + + private static Pair<Transform, String> convertAttr(Expression expr) { + Transform type = TRANSFORMS.get(expr.getClass()); + if (type == Transform.IDENTITY) { + Attribute attr = (Attribute) expr; + return Pair.of(type, attr.name()); + } else if (expr instanceof Cast) { + Cast cast = (Cast) expr; + if (DateType$.MODULE$.sameType(cast.dataType()) && cast.child() instanceof Attribute) { + Attribute attr = (Attribute) cast.child(); + return Pair.of(Transform.DAY, attr.name()); + } + } else if (expr instanceof ParseToDate) { + ParseToDate toDate = (ParseToDate) expr; + if (toDate.left() instanceof Attribute) { + Attribute attr = (Attribute) toDate.left(); + return Pair.of(Transform.DAY, attr.name()); + } + } else if (expr instanceof UnaryExpression) { + UnaryExpression func = (UnaryExpression) expr; + if (func.child() instanceof Attribute) { + Attribute attr = (Attribute) func.child(); + return Pair.of(type, attr.name()); + } + } + return null; + } + + private static long yearToTimestampMicros(int year) { + return ChronoUnit.MICROS.between(EPOCH, + LocalDateTime.of(year, 1, 1, 0, 0).atOffset(ZoneOffset.UTC)); + } + + private static long dayToTimestampMicros(int daysFromEpoch) { + return ChronoUnit.MICROS.between(EPOCH, + EPOCH_DAY.plusDays(daysFromEpoch).atStartOfDay().atOffset(ZoneOffset.UTC)); + } + + private static org.apache.iceberg.expressions.Literal<Long> tsLiteral(long timestampMicros) { + return org.apache.iceberg.expressions.Literal + .of(timestampMicros) + .to(TimestampType.withoutZone()); + } + + private static org.apache.iceberg.expressions.Expression filter( + Operation op, String name, int value, Function<Integer, Long> startTsMicros) { + switch (op) { + case LT: + return predicate(Operation.LT, name, tsLiteral(startTsMicros.apply(value))); + case LT_EQ: + return predicate(Operation.LT, name, tsLiteral(startTsMicros.apply(value + 1))); + case GT: + return predicate(Operation.GT_EQ, name, tsLiteral(startTsMicros.apply(value + 1))); + case GT_EQ: + return predicate(Operation.GT_EQ, name, tsLiteral(startTsMicros.apply(value))); + case EQ: + return and( + predicate(Operation.GT_EQ, name, tsLiteral(startTsMicros.apply(value))), + predicate(Operation.LT, name, tsLiteral(startTsMicros.apply(value + 1))) + ); + case NOT_EQ: + return or( + predicate(Operation.GT_EQ, name, tsLiteral(startTsMicros.apply(value + 1))), + predicate(Operation.LT, name, tsLiteral(startTsMicros.apply(value))) + ); + case IN: + case NOT_IN: + default: + throw new IllegalArgumentException("Cannot convert operation to year filter: " + op); + } + } + + private enum Transform { Review comment: I have referred this code from https://github.com/Netflix/iceberg/blob/master/spark/src/main/java/com/netflix/iceberg/spark/SparkExpressions.java Currently testcases are passing as the filter conversion is success. I know that we can add some more transform here. But for the case of rewrite data file procedure , the filters will be always `identity` transform as it comes from spark ? So, no need of other transform ? -- 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]
