rdblue commented on code in PR #12496: URL: https://github.com/apache/iceberg/pull/12496#discussion_r2007936907
########## parquet/src/main/java/org/apache/iceberg/parquet/ParquetMetrics.java: ########## @@ -0,0 +1,613 @@ +/* + * + * * 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.parquet; + +import java.nio.ByteBuffer; +import java.util.Comparator; +import java.util.Deque; +import java.util.List; +import java.util.Map; +import java.util.Set; +import java.util.function.Function; +import java.util.stream.Collectors; +import java.util.stream.Stream; +import org.apache.iceberg.FieldMetrics; +import org.apache.iceberg.Metrics; +import org.apache.iceberg.MetricsConfig; +import org.apache.iceberg.MetricsModes; +import org.apache.iceberg.MetricsUtil; +import org.apache.iceberg.Schema; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Iterables; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Maps; +import org.apache.iceberg.relocated.com.google.common.collect.Multimap; +import org.apache.iceberg.relocated.com.google.common.collect.Multimaps; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.relocated.com.google.common.collect.Streams; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.types.Conversions; +import org.apache.iceberg.types.Types; +import org.apache.iceberg.util.BinaryUtil; +import org.apache.iceberg.util.NaNUtil; +import org.apache.iceberg.util.UnicodeUtil; +import org.apache.iceberg.variants.PhysicalType; +import org.apache.iceberg.variants.ShreddedObject; +import org.apache.iceberg.variants.VariantMetadata; +import org.apache.iceberg.variants.VariantValue; +import org.apache.iceberg.variants.Variants; +import org.apache.parquet.column.statistics.Statistics; +import org.apache.parquet.hadoop.metadata.BlockMetaData; +import org.apache.parquet.hadoop.metadata.ColumnChunkMetaData; +import org.apache.parquet.hadoop.metadata.ColumnPath; +import org.apache.parquet.hadoop.metadata.ParquetMetadata; +import org.apache.parquet.schema.GroupType; +import org.apache.parquet.schema.MessageType; +import org.apache.parquet.schema.PrimitiveType; +import org.apache.parquet.schema.Type; + +class ParquetMetrics { + private ParquetMetrics() {} + + static Metrics metrics( + Schema schema, + MessageType type, + MetricsConfig metricsConfig, + ParquetMetadata metadata, + Stream<FieldMetrics<?>> fields) { + long rowCount = 0L; + Map<Integer, Long> columnSizes = Maps.newHashMap(); + Multimap<ColumnPath, ColumnChunkMetaData> columns = + Multimaps.newMultimap(Maps.newHashMap(), Lists::newArrayList); + for (BlockMetaData block : metadata.getBlocks()) { + rowCount += block.getRowCount(); + for (ColumnChunkMetaData column : block.getColumns()) { + Type.ID id = + type.getColumnDescription(column.getPath().toArray()).getPrimitiveType().getId(); + if (null == id) { + continue; + } + + int fieldId = id.intValue(); + MetricsModes.MetricsMode mode = MetricsUtil.metricsMode(schema, metricsConfig, fieldId); + if (mode != MetricsModes.None.get()) { + columns.put(column.getPath(), column); + columnSizes.put(fieldId, columnSizes.getOrDefault(fieldId, 0L) + column.getTotalSize()); + } + } + } + + Map<Integer, FieldMetrics<?>> metricsById = + fields.collect(Collectors.toMap(FieldMetrics::id, Function.identity())); + + Iterable<FieldMetrics<ByteBuffer>> results = + TypeWithSchemaVisitor.visit( + schema.asStruct(), + type, + new MetricsVisitor(schema, metricsConfig, metricsById, columns)); + + Map<Integer, Long> valueCounts = Maps.newHashMap(); + Map<Integer, Long> nullValueCounts = Maps.newHashMap(); + Map<Integer, Long> nanValueCounts = Maps.newHashMap(); + Map<Integer, ByteBuffer> lowerBounds = Maps.newHashMap(); + Map<Integer, ByteBuffer> upperBounds = Maps.newHashMap(); + + for (FieldMetrics<ByteBuffer> metrics : results) { + int id = metrics.id(); + if (metrics.valueCount() >= 0) { + valueCounts.put(id, metrics.valueCount()); + } + + if (metrics.nullValueCount() >= 0) { + nullValueCounts.put(id, metrics.nullValueCount()); + } + + if (metrics.nanValueCount() >= 0) { + nanValueCounts.put(id, metrics.nanValueCount()); + } + + if (metrics.hasBounds()) { + lowerBounds.put(id, metrics.lowerBound()); + upperBounds.put(id, metrics.upperBound()); + } + } + + return new Metrics( + rowCount, + columnSizes, + valueCounts, + nullValueCounts, + nanValueCounts, + lowerBounds, + upperBounds); + } + + private static class MetricsVisitor + extends TypeWithSchemaVisitor<Iterable<FieldMetrics<ByteBuffer>>> { + private final Schema schema; + private final MetricsConfig metricsConfig; + private final Map<Integer, FieldMetrics<?>> metricsById; + private final Multimap<ColumnPath, ColumnChunkMetaData> columns; + + private MetricsVisitor( + Schema schema, + MetricsConfig metricsConfig, + Map<Integer, FieldMetrics<?>> metricsById, + Multimap<ColumnPath, ColumnChunkMetaData> columns) { + this.schema = schema; + this.metricsConfig = metricsConfig; + this.metricsById = metricsById; + this.columns = columns; + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> message( + Types.StructType iStruct, + MessageType message, + List<Iterable<FieldMetrics<ByteBuffer>>> fieldResults) { + return Iterables.concat(fieldResults); + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> struct( + Types.StructType iStruct, + GroupType struct, + List<Iterable<FieldMetrics<ByteBuffer>>> fieldResults) { + return Iterables.concat(fieldResults); + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> list( + Types.ListType iList, GroupType array, Iterable<FieldMetrics<ByteBuffer>> elementResults) { + // remove lower and upper bounds for repeated fields + return ImmutableList.of(); + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> map( + Types.MapType iMap, + GroupType map, + Iterable<FieldMetrics<ByteBuffer>> keyResults, + Iterable<FieldMetrics<ByteBuffer>> valueResults) { + // repeated fields are not currently supported + return ImmutableList.of(); + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> primitive( + org.apache.iceberg.types.Type.PrimitiveType iPrimitive, PrimitiveType primitive) { + Type.ID id = primitive.getId(); + if (null == id) { + return ImmutableList.of(); + } + int fieldId = id.intValue(); + + MetricsModes.MetricsMode mode = MetricsUtil.metricsMode(schema, metricsConfig, fieldId); + if (mode == MetricsModes.None.get()) { + return ImmutableList.of(); + } + + int length = truncateLength(mode); + + FieldMetrics<ByteBuffer> metrics = metricsFromFieldMetrics(fieldId, iPrimitive, length); + if (metrics != null) { + return ImmutableList.of(metrics); + } + + metrics = metricsFromFooter(fieldId, iPrimitive, primitive, length); + if (metrics != null) { + return ImmutableList.of(metrics); + } + + return ImmutableList.of(); + } + + private FieldMetrics<ByteBuffer> metricsFromFieldMetrics( + int fieldId, org.apache.iceberg.types.Type.PrimitiveType icebergType, int truncateLength) { + FieldMetrics<?> fieldMetrics = metricsById.get(fieldId); + if (null == fieldMetrics) { + return null; + } else if (truncateLength <= 0) { + return new FieldMetrics<>( + fieldMetrics.id(), + fieldMetrics.valueCount(), + fieldMetrics.nullValueCount(), + fieldMetrics.nanValueCount()); + } else { + Object lowerBound = + truncateLowerBound(icebergType, fieldMetrics.lowerBound(), truncateLength); + Object upperBound = + truncateUpperBound(icebergType, fieldMetrics.upperBound(), truncateLength); + ByteBuffer lower = Conversions.toByteBuffer(icebergType, lowerBound); + ByteBuffer upper = Conversions.toByteBuffer(icebergType, upperBound); + return new FieldMetrics<>( + fieldMetrics.id(), + fieldMetrics.valueCount(), + fieldMetrics.nullValueCount(), + fieldMetrics.nanValueCount(), + lower, + upper); + } + } + + private FieldMetrics<ByteBuffer> metricsFromFooter( + int fieldId, + org.apache.iceberg.types.Type.PrimitiveType icebergType, + PrimitiveType primitive, + int truncateLength) { + if (primitive.getPrimitiveTypeName() == PrimitiveType.PrimitiveTypeName.INT96) { + return null; + } else if (truncateLength <= 0) { + return counts(fieldId); + } else { + return bounds(fieldId, icebergType, primitive, truncateLength); + } + } + + private FieldMetrics<ByteBuffer> counts(int fieldId) { + ColumnPath path = ColumnPath.get(currentPath()); + long valueCount = 0; + long nullCount = 0; + + for (ColumnChunkMetaData column : columns.get(path)) { + Statistics<?> stats = column.getStatistics(); + if (stats == null || stats.isEmpty()) { + return null; + } + + nullCount += stats.getNumNulls(); + valueCount += column.getValueCount(); + } + + return new FieldMetrics<>(fieldId, valueCount, nullCount); + } + + private <T> FieldMetrics<ByteBuffer> bounds( + int fieldId, + org.apache.iceberg.types.Type.PrimitiveType icebergType, + PrimitiveType primitive, + int truncateLength) { + if (icebergType == null) { + return null; + } + + ColumnPath path = ColumnPath.get(currentPath()); + Comparator<T> comparator = Comparators.forType(icebergType); + long valueCount = 0; + long nullCount = 0; + T lowerBound = null; + T upperBound = null; + + for (ColumnChunkMetaData column : columns.get(path)) { + Statistics<?> stats = column.getStatistics(); + if (stats == null || stats.isEmpty()) { + return null; + } + + nullCount += stats.getNumNulls(); + valueCount += column.getValueCount(); + + if (stats.hasNonNullValue()) { + T chunkMin = + ParquetConversions.convertValue(icebergType, primitive, stats.genericGetMin()); + if (lowerBound == null || comparator.compare(chunkMin, lowerBound) < 0) { + lowerBound = chunkMin; + } + + T chunkMax = + ParquetConversions.convertValue(icebergType, primitive, stats.genericGetMax()); + if (upperBound == null || comparator.compare(chunkMax, upperBound) > 0) { + upperBound = chunkMax; + } + } + } + + if (NaNUtil.isNaN(lowerBound) || NaNUtil.isNaN(upperBound)) { + return new FieldMetrics<>(fieldId, valueCount, nullCount); + } + + lowerBound = truncateLowerBound(icebergType, lowerBound, truncateLength); + upperBound = truncateUpperBound(icebergType, upperBound, truncateLength); + + ByteBuffer lower = Conversions.toByteBuffer(icebergType, lowerBound); + ByteBuffer upper = Conversions.toByteBuffer(icebergType, upperBound); + + return new FieldMetrics<>(fieldId, valueCount, nullCount, lower, upper); + } + + @Override + public Iterable<FieldMetrics<ByteBuffer>> variant( + Types.VariantType iVariant, GroupType variant, Iterable<FieldMetrics<ByteBuffer>> ignored) { + Type.ID id = variant.getId(); + if (null == id) { + return ImmutableList.of(); + } + int fieldId = id.intValue(); + + MetricsModes.MetricsMode mode = MetricsUtil.metricsMode(schema, metricsConfig, fieldId); + if (mode == MetricsModes.None.get()) { + return ImmutableList.of(); + } + + List<ParquetVariantUtil.VariantMetrics> results = + Lists.newArrayList( + ParquetVariantVisitor.visit(variant, new MetricsVariantVisitor(currentPath()))); + + if (results.size() <= 1) { + return ImmutableList.of(); + } + + ParquetVariantUtil.VariantMetrics metadataCounts = results.get(0); + if (mode == MetricsModes.Counts.get()) { + return ImmutableList.of( + new FieldMetrics<>(fieldId, metadataCounts.valueCount(), metadataCounts.nullCount())); + } + + Set<String> fieldNames = Sets.newTreeSet(); + for (ParquetVariantUtil.VariantMetrics result : results.subList(1, results.size())) { + fieldNames.add(result.fieldName()); + } + + VariantMetadata metadata = Variants.metadata(fieldNames); + ShreddedObject lowerBounds = Variants.object(metadata); + ShreddedObject upperBounds = Variants.object(metadata); + for (ParquetVariantUtil.VariantMetrics result : results.subList(1, results.size())) { + String fieldName = result.fieldName(); + lowerBounds.put(fieldName, result.lowerBound()); + upperBounds.put(fieldName, result.upperBound()); + } + + return ImmutableList.of( + new FieldMetrics<>( Review Comment: Right now, I just wanted to get the basics working so I went with the simplest implementation. We should definitely revisit this and discuss how to configure metrics collection. That said, I think the mode isn't the right problem to solve. For the mode, we don't keep counts other than the top-level value and null count for the variant itself. That leaves only how to handle lower and upper bounds, where we know that `truncate` is the right config and 16 is a reasonable length default. At that point, the only question is whether we want to hard-code it, pass through the mode for the variant column to use a configurable length for all sub-fields, or if we want to have a truncate length for each field individually. The bigger problem is which fields to collect metrics for. Restricting the fields to just the ones that are shredded is a good heuristic because we don't expect types to be uniform for other fields, and a value of another type will prevent the field's bounds from being stored. Even then, there could be quite a few fields and that will make the lower and upper bound payloads large. We may want to further restrict the number of fields, but for now I think the reasonable path forward is to use the shredded fields. Then we can see if we want to change it once we tackle the problem of how we determine the fields to shred. -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
