Avro Parquet

The Avro Parquet connector provides an Akka Stream Source, Sink and Flow for push and pull data to and from Parquet files.

For more information about Apache Parquet please visit the official documentation.

Project Info: Alpakka Avro Parquet
Artifact
com.lightbend.akka
akka-stream-alpakka-avroparquet
3.0.4
JDK versions
Adopt OpenJDK 8
Adopt OpenJDK 11
Scala versions2.12.11, 2.13.3
JPMS module nameakka.stream.alpakka.avroparquet
License
Readiness level
Since 1.0-M1, 2018-11-06
Home pagehttps://doc.akka.io/docs/alpakka/current
API documentation
Forums
Release notesGitHub releases
IssuesGithub issues
Sourceshttps://github.com/akka/alpakka

Artifacts

sbt
val AkkaVersion = "2.6.14"
libraryDependencies ++= Seq(
  "com.lightbend.akka" %% "akka-stream-alpakka-avroparquet" % "3.0.4",
  "com.typesafe.akka" %% "akka-stream" % AkkaVersion
)
Maven
<properties>
  <akka.version>2.6.14</akka.version>
  <scala.binary.version>2.12</scala.binary.version>
</properties>
<dependencies>
  <dependency>
    <groupId>com.lightbend.akka</groupId>
    <artifactId>akka-stream-alpakka-avroparquet_${scala.binary.version}</artifactId>
    <version>3.0.4</version>
  </dependency>
  <dependency>
    <groupId>com.typesafe.akka</groupId>
    <artifactId>akka-stream_${scala.binary.version}</artifactId>
    <version>${akka.version}</version>
  </dependency>
</dependencies>
Gradle
def versions = [
  AkkaVersion: "2.6.14",
  ScalaBinary: "2.12"
]
dependencies {
  implementation "com.lightbend.akka:akka-stream-alpakka-avroparquet_${versions.ScalaBinary}:3.0.4"
  implementation "com.typesafe.akka:akka-stream_${versions.ScalaBinary}:${versions.AkkaVersion}"
}

The table below shows direct dependencies of this module and the second tab shows all libraries it depends on transitively.

Direct dependencies
OrganizationArtifactVersion
com.typesafe.akkaakka-stream_2.122.6.14
org.apache.parquetparquet-avro1.10.1
org.scala-langscala-library2.12.11
Dependency tree
com.typesafe.akka    akka-stream_2.12    2.6.14    Apache-2.0
    com.typesafe.akka    akka-actor_2.12    2.6.14    Apache-2.0
        com.typesafe    config    1.4.0    Apache-2.0
        org.scala-lang.modules    scala-java8-compat_2.12    0.8.0    BSD 3-clause
            org.scala-lang    scala-library    2.12.11    Apache-2.0
        org.scala-lang    scala-library    2.12.11    Apache-2.0
    com.typesafe.akka    akka-protobuf-v3_2.12    2.6.14    Apache-2.0
    com.typesafe    ssl-config-core_2.12    0.4.2    Apache-2.0
        com.typesafe    config    1.4.0    Apache-2.0
        org.scala-lang.modules    scala-parser-combinators_2.12    1.1.2    Apache-2.0
            org.scala-lang    scala-library    2.12.11    Apache-2.0
        org.scala-lang    scala-library    2.12.11    Apache-2.0
    org.reactivestreams    reactive-streams    1.0.3    CC0
    org.scala-lang    scala-library    2.12.11    Apache-2.0
org.apache.parquet    parquet-avro    1.10.1
    it.unimi.dsi    fastutil    7.0.13    Apache License, Version 2.0
    org.apache.avro    avro    1.8.2
        com.thoughtworks.paranamer    paranamer    2.7
        org.apache.commons    commons-compress    1.8.1
        org.codehaus.jackson    jackson-core-asl    1.9.13    The Apache Software License, Version 2.0
        org.codehaus.jackson    jackson-mapper-asl    1.9.13    The Apache Software License, Version 2.0
            org.codehaus.jackson    jackson-core-asl    1.9.13    The Apache Software License, Version 2.0
        org.slf4j    slf4j-api    1.7.22
        org.tukaani    xz    1.5    Public Domain
        org.xerial.snappy    snappy-java    1.1.2.6    The Apache Software License, Version 2.0
    org.apache.parquet    parquet-column    1.10.1
        commons-codec    commons-codec    1.10
        org.apache.parquet    parquet-common    1.10.1
            org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
                org.slf4j    slf4j-api    1.7.22
            org.slf4j    slf4j-api    1.7.22
        org.apache.parquet    parquet-encoding    1.10.1
            commons-codec    commons-codec    1.10
            org.apache.parquet    parquet-common    1.10.1
                org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
                    org.slf4j    slf4j-api    1.7.22
                org.slf4j    slf4j-api    1.7.22
    org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
        org.slf4j    slf4j-api    1.7.22
    org.apache.parquet    parquet-hadoop    1.10.1
        commons-pool    commons-pool    1.6
        org.apache.parquet    parquet-column    1.10.1
            commons-codec    commons-codec    1.10
            org.apache.parquet    parquet-common    1.10.1
                org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
                    org.slf4j    slf4j-api    1.7.22
                org.slf4j    slf4j-api    1.7.22
            org.apache.parquet    parquet-encoding    1.10.1
                commons-codec    commons-codec    1.10
                org.apache.parquet    parquet-common    1.10.1
                    org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
                        org.slf4j    slf4j-api    1.7.22
                    org.slf4j    slf4j-api    1.7.22
        org.apache.parquet    parquet-format    2.4.0    The Apache Software License, Version 2.0
            org.slf4j    slf4j-api    1.7.22
        org.apache.parquet    parquet-jackson    1.10.1
        org.codehaus.jackson    jackson-core-asl    1.9.13    The Apache Software License, Version 2.0
        org.codehaus.jackson    jackson-mapper-asl    1.9.13    The Apache Software License, Version 2.0
            org.codehaus.jackson    jackson-core-asl    1.9.13    The Apache Software License, Version 2.0
        org.xerial.snappy    snappy-java    1.1.2.6    The Apache Software License, Version 2.0
org.scala-lang    scala-library    2.12.11    Apache-2.0

Source Initiation

Sometimes it might be useful to use a Parquet file as stream Source. For this we will need to create an AvroParquetReader instance which will produce records as subtypes of GenericRecord, the Avro record’s abstract representation.

Scala
sourceimport org.apache.hadoop.conf.Configuration
import org.apache.parquet.avro.AvroReadSupport

val conf: Configuration = new Configuration()
conf.setBoolean(AvroReadSupport.AVRO_COMPATIBILITY, true)
val reader: ParquetReader[GenericRecord] =
  AvroParquetReader.builder[GenericRecord](HadoopInputFile.fromPath(new Path(file), conf)).withConf(conf).build()
Java
sourceimport org.apache.parquet.hadoop.ParquetReader;
import org.apache.avro.generic.GenericRecord;
import org.apache.hadoop.fs.Path;
import org.apache.avro.Schema;
import akka.stream.javadsl.Source;
import org.apache.parquet.avro.AvroParquetReader;

Configuration conf = new Configuration();

ParquetReader<GenericRecord> reader =
    AvroParquetReader.<GenericRecord>builder(
            HadoopInputFile.fromPath(new Path("./test.parquet"), conf))
        .disableCompatibility()
        .build();

After that, you can create the Parquet Source from the initialisation of AvroParquetReader. This object requires an instance of a org.apache.parquet.hadoop.ParquetReader typed by a subtype of GenericRecord.

Scala
sourceval source: Source[GenericRecord, NotUsed] = AvroParquetSource(reader)
val source: Source[GenericRecord, NotUsed] = AvroParquetSource(reader)
Java
sourceSource<GenericRecord, NotUsed> source = AvroParquetSource.create(reader);

Sink Initiation

On the other hand, you can use AvroParquetWriter as the Akka Streams Sink implementation for writing to Parquet. In that case, its initialisation would require an instance of org.apache.parquet.hadoop.ParquetWriter. It will also expect any subtype of GenericRecord to be passed.

Scala
sourceimport com.sksamuel.avro4s.Record
import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.fs.Path
import org.apache.parquet.avro.AvroReadSupport

val file: String = "./sample/path/test.parquet"
val conf: Configuration = new Configuration()
conf.setBoolean(AvroReadSupport.AVRO_COMPATIBILITY, true)
val writer: ParquetWriter[Record] =
  AvroParquetWriter.builder[Record](new Path(file)).withConf(conf).withSchema(schema).build()
Java
sourceimport org.apache.parquet.hadoop.ParquetWriter;
import org.apache.avro.Schema;
import org.apache.avro.generic.GenericRecord;
import org.apache.avro.generic.GenericRecordBuilder;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.parquet.hadoop.util.HadoopInputFile;

Configuration conf = new Configuration();
conf.setBoolean(AvroReadSupport.AVRO_COMPATIBILITY, true);
ParquetWriter<GenericRecord> writer =
    AvroParquetWriter.<GenericRecord>builder(new Path(file))
        .withConf(conf)
        .withWriteMode(ParquetFileWriter.Mode.OVERWRITE)
        .withSchema(schema)
        .build();

After that, the AvroParquet Sink can already be used.

The below Scala example demonstrates that any subtype of GenericRecord can be passed to the stream. In this case the one used is com.sksamuel.avro4s.Record, which it implements the GenericRecord Avro interface. See Avro4s or Avrohugger for other ways of generating these classes.

Scala
sourceval records: List[Record] = documents.map(RecordFormat[Document].to(_))
val source: Source[Record, NotUsed] = Source(records)
val result: Future[Done] = source
  .runWith(AvroParquetSink(writer))
Java
sourceSink<GenericRecord, CompletionStage<Done>> sink = AvroParquetSink.create(writer);

Flow Initiation

The representation of a ParquetWriter as a Flow is also available to use as a streams flow stage, in which as well as the other representations, it will expect subtypes of the Parquet GenericRecord type to be passed. As a result, it writes into a Parquet file and returns the same GenericRecords. Such a Flow stage can be easily created by using the AvroParquetFlow and providing an AvroParquetWriter instance as a parameter.

Scala
sourceval records: List[GenericRecord]
val source: Source[GenericRecord, NotUsed] = Source(records)
val avroParquet: Flow[GenericRecord, GenericRecord, NotUsed] = AvroParquetFlow(writer)
val result =
  source
    .via(avroParquet)
    .runWith(Sink.seq)
This is all the preparation that we are going to need.
Java
sourceParquetWriter<GenericRecord> writer =
    AvroParquetWriter.<GenericRecord>builder(new Path("./test.parquet"))
        .withConf(conf)
        .withSchema(schema)
        .build();

Flow<GenericRecord, GenericRecord, NotUsed> flow = AvroParquetFlow.create(writer);

source.via(flow).runWith(Sink.ignore(), system);

Running the example code

The code in this guide is part of the runnable tests of this project. You are welcome to edit the code and run it in sbt.

Scala
sbt
> avroparquet/test
Found an error in this documentation? The source code for this page can be found here. Please feel free to edit and contribute a pull request.