Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Kernel] Default Parquet reader implementation #1846

Closed
wants to merge 9 commits into from
Closed
Show file tree
Hide file tree
Changes from 3 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 1 addition & 5 deletions kernel/build.sbt
Original file line number Diff line number Diff line change
Expand Up @@ -78,17 +78,13 @@ lazy val kernelDefault = (project in file("kernel-default"))
scalaStyleSettings,
releaseSettings,
libraryDependencies ++= Seq(
"org.apache.hadoop" % "hadoop-client-api" % hadoopVersion, // Configuration, Path
"org.apache.hadoop" % "hadoop-client-runtime" % hadoopVersion, // Configuration, Path
"io.delta" % "delta-storage" % deltaStorageVersion, // LogStore
"com.fasterxml.jackson.core" % "jackson-databind" % "2.13.5", // ObjectMapper
"org.apache.parquet" % "parquet-hadoop" % "1.12.3",

"org.scalatest" %% "scalatest" % scalaTestVersion % "test",
"io.delta" %% "delta-core" % deltaSparkVersion % "test",
"org.apache.spark" %% "spark-sql" % sparkVersion % "test", // SparkSession
"org.apache.spark" %% "spark-sql" % sparkVersion % "test" classifier "tests",
"org.apache.spark" %% "spark-core" % sparkVersion % "test" classifier "tests",
"org.apache.spark" %% "spark-catalyst" % sparkVersion % "test" classifier "tests",
"junit" % "junit" % "4.11" % "test",
"com.novocode" % "junit-interface" % "0.11" % "test"
)
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,8 @@
import java.util.function.Consumer;
import java.util.function.Function;

public interface CloseableIterator<T> extends Iterator<T>, Closeable {
public interface CloseableIterator<T> extends Iterator<T>, Closeable
{
default <U> CloseableIterator<U> map(Function<T, U> mapper) {
CloseableIterator<T> delegate = this;
return new CloseableIterator<U>() {
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,187 @@
/*
* Copyright (2023) The Delta Lake Project Authors.
*
* Licensed 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 io.delta.kernel;

import java.util.ArrayList;
import java.util.List;
import org.apache.parquet.schema.GroupType;
import org.apache.parquet.schema.MessageType;
import org.apache.parquet.schema.Type;

import io.delta.kernel.types.DataType;
import io.delta.kernel.types.StructField;
import io.delta.kernel.types.StructType;

public class DefaultKernelUtils
{
private DefaultKernelUtils() {}

/**
* Given the file schema in Parquet file and selected columns by Delta, return
* a subschema of the file schema.
*
* @param fileSchema
* @param deltaType
* @return
*/
public static final MessageType pruneSchema(
MessageType fileSchema, // parquet
StructType deltaType) // delta-core
{
return deltaType.fields().stream()
.map(column -> {
Type type = findStructField(fileSchema, column);
if (type == null) {
return null;
}
Type prunedSubfields = pruneSubfields(type, column.getDataType());
return new MessageType(column.getName(), prunedSubfields);
})
.filter(type -> type != null)
.reduce(MessageType::union)
.get();
}

private static Type findStructField(MessageType fileSchema, StructField column)
{
// TODO: Need a way to search by id once we start supporting column mapping `id` mode.
final String columnName = column.getName();
if (fileSchema.containsField(columnName)) {
return fileSchema.getType(columnName);
}
// Parquet is case-sensitive, but hive is not. all hive columns get converted to lowercase
// check for direct match above but if no match found, try case-insensitive match
for (org.apache.parquet.schema.Type type : fileSchema.getFields()) {
if (type.getName().equalsIgnoreCase(columnName)) {
return type;
}
}

return null;
}

private static Type pruneSubfields(Type type, DataType deltaDatatype)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems like this and pruneSchema can use the same code if we generalize it for GroupType?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cleaned up a bit, but not clean to merge them. MessageType and Group type needs two different inputs. For now keeping them separate. It only adds extra 5lines of code.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This only prunes the 2nd level of nesting right?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can't we still factor out the shared code?

private static List<Type> pruneFields(GroupType type, DataType deltaDataType) {
        // prune fields including nested pruning like in pruneSchema
        return deltaType.fields().stream()
            .map(column -> {
                Type type = findSubFieldType(type, column);
                if (type != null && column instanceof StructType) {
                       return type.withNewFields(pruneFields(type, column);
                } else {
                      return type;
                }
            })
            .filter(Objects::nonNull)
            .collect(Collectors.toList());
}

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You need the MessageType for the pruneSchema.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

MessageType extends GroupType through right? So pruneSchema can just call a helper. Not a big deal though, can we just either fix the pruning for nested fields or add a note that nested schema pruning is only 2nd level?

vkorukanti marked this conversation as resolved.
Show resolved Hide resolved
{
if (!(deltaDatatype instanceof StructType)) {
// there is no pruning for non-struct types
return type;
}

GroupType groupType = (GroupType) type;
StructType deltaStructType = (StructType) deltaDatatype;
List<Type> newParquetSubFields = new ArrayList<>();
for (StructField subField : deltaStructType.fields()) {
String subFieldName = subField.getName();
Type parquetSubFieldType = groupType.getType(subFieldName);
if (parquetSubFieldType == null) {
for (org.apache.parquet.schema.Type typeTemp : groupType.getFields()) {
if (typeTemp.getName().equalsIgnoreCase(subFieldName)) {
parquetSubFieldType = type;
}
}
}
newParquetSubFields.add(parquetSubFieldType);
}
return groupType.withNewFields(newParquetSubFields);
}

/**
* Precondition-style validation that throws {@link IllegalArgumentException}.
*
* @param isValid {@code true} if valid, {@code false} if an exception should be thrown
* @throws IllegalArgumentException if {@code isValid} is false
*/
public static void checkArgument(boolean isValid)
throws IllegalArgumentException
{
if (!isValid) {
throw new IllegalArgumentException();
}
}

/**
* Precondition-style validation that throws {@link IllegalArgumentException}.
*
* @param isValid {@code true} if valid, {@code false} if an exception should be thrown
* @param message A String message for the exception.
* @throws IllegalArgumentException if {@code isValid} is false
*/
public static void checkArgument(boolean isValid, String message)
throws IllegalArgumentException
{
if (!isValid) {
throw new IllegalArgumentException(message);
}
}

/**
* Precondition-style validation that throws {@link IllegalArgumentException}.
*
* @param isValid {@code true} if valid, {@code false} if an exception should be thrown
* @param message A String message for the exception.
* @param args Objects used to fill in {@code %s} placeholders in the message
* @throws IllegalArgumentException if {@code isValid} is false
*/
public static void checkArgument(boolean isValid, String message, Object... args)
throws IllegalArgumentException
{
if (!isValid) {
throw new IllegalArgumentException(
String.format(String.valueOf(message), args));
}
}

/**
* Precondition-style validation that throws {@link IllegalStateException}.
*
* @param isValid {@code true} if valid, {@code false} if an exception should be thrown
* @param message A String message for the exception.
* @throws IllegalStateException if {@code isValid} is false
*/
public static void checkState(boolean isValid, String message)
throws IllegalStateException
{
if (!isValid) {
throw new IllegalStateException(message);
}
}

/**
* Search for the Parquet type for in the {@code groupType} for the field equilant to
* {@code field}.
*
* @param groupType Parquet group type coming from the file schema.
* @param field Sub field given as Delta Kernel's {@link StructField}
* @return {@link Type} of the Parquet field. Returns {@code null}, if not found.
*/
public static Type findFieldType(GroupType groupType, StructField field)
vkorukanti marked this conversation as resolved.
Show resolved Hide resolved
{
// TODO: Need a way to search by id once we start supporting column mapping `id` mode.
final String columnName = field.getName();
if (groupType.containsField(columnName)) {
return groupType.getType(columnName);
}
// Parquet is case-sensitive, but hive is not. all hive columns get converted to lowercase
// check for direct match above but if no match found, try case-insensitive match
for (org.apache.parquet.schema.Type type : groupType.getFields()) {
if (type.getName().equalsIgnoreCase(columnName)) {
return type;
}
}

return null;
}
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,133 @@
/*
* Copyright (2023) The Delta Lake Project Authors.
*
* Licensed 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 io.delta.kernel.client;

import java.io.IOException;
import org.apache.hadoop.conf.Configuration;

import io.delta.kernel.data.ColumnarBatch;
import io.delta.kernel.data.FileDataReadResult;
import io.delta.kernel.data.Row;
import io.delta.kernel.expressions.Expression;
import io.delta.kernel.fs.FileStatus;
import io.delta.kernel.parquet.ParquetBatchReader;
import io.delta.kernel.types.StructType;
import io.delta.kernel.utils.CloseableIterator;
import io.delta.kernel.utils.Utils;

public class DefaultParquetHandler
implements ParquetHandler
{
private final Configuration hadoopConf;

public DefaultParquetHandler(Configuration hadoopConf)
{
this.hadoopConf = hadoopConf;
}

@Override
public CloseableIterator<FileReadContext> contextualizeFileReads(
CloseableIterator<Row> fileIter,
Expression predicate)
{
return new CloseableIterator<FileReadContext>()
{
@Override
public void close()
throws IOException
{
fileIter.close();
}

@Override
public boolean hasNext()
{
return fileIter.hasNext();
}

@Override
public FileReadContext next()
{
return () -> fileIter.next();
vkorukanti marked this conversation as resolved.
Show resolved Hide resolved
}
};
}

@Override
public CloseableIterator<FileDataReadResult> readParquetFiles(
CloseableIterator<FileReadContext> fileIter,
StructType physicalSchema) throws IOException
{
return new CloseableIterator<FileDataReadResult>()
{
private FileReadContext currentFile;
private CloseableIterator<ColumnarBatch> currentFileReader;

@Override
public void close()
throws IOException
{
if (currentFileReader != null) {
currentFileReader.close();
}

fileIter.close();
// TODO: implement safe close of multiple closeables.
}

@Override
public boolean hasNext()
{
// There is no file in reading or the current file being read has no more data
// initialize the next file reader or return false if there are no more files to
// read.
if (currentFileReader == null || !currentFileReader.hasNext()) {
if (fileIter.hasNext()) {
currentFile = fileIter.next();
FileStatus fileStatus = Utils.getFileStatus(currentFile.getScanFileRow());
ParquetBatchReader batchReader = new ParquetBatchReader(hadoopConf);
vkorukanti marked this conversation as resolved.
Show resolved Hide resolved
currentFileReader = batchReader.read(fileStatus.getPath(), physicalSchema);
}
else {
return false;
}
}

return currentFileReader.hasNext();
}

@Override
public FileDataReadResult next()
{
final ColumnarBatch data = currentFileReader.next();
return new FileDataReadResult()
{
@Override
public ColumnarBatch getData()
{
return data;
}

@Override
public Row getScanFileRow()
{
return currentFile.getScanFileRow();
}
};
}
};
}
}
Loading