[SPARK-46679] Encoders with multiple inheritance - Key not found: T

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 3.4.2, 3.5.0, 4.0.0
• Component/s: SQL
• Labels: pull-request-available

Language

Java Scala

• Assignee: Unassigned
• Reporter: Andoni Teso
• Created: 11/Jan/24 10:56
• Updated: 24/Oct/24 18:50

Description

Since version 3.4, I've been experiencing the following error when using encoders.

Exception in thread "main" java.util.NoSuchElementException: key not found: T
    at scala.collection.immutable.Map$Map1.apply(Map.scala:163)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.encoderFor(JavaTypeInference.scala:121)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.$anonfun$encoderFor$1(JavaTypeInference.scala:140)
    at scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:286)
    at scala.collection.IndexedSeqOptimized.foreach(IndexedSeqOptimized.scala:36)
    at scala.collection.IndexedSeqOptimized.foreach$(IndexedSeqOptimized.scala:33)
    at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:198)
    at scala.collection.TraversableLike.map(TraversableLike.scala:286)
    at scala.collection.TraversableLike.map$(TraversableLike.scala:279)
    at scala.collection.mutable.ArrayOps$ofRef.map(ArrayOps.scala:198)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.encoderFor(JavaTypeInference.scala:138)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.$anonfun$encoderFor$1(JavaTypeInference.scala:140)
    at scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:286)
    at scala.collection.IndexedSeqOptimized.foreach(IndexedSeqOptimized.scala:36)
    at scala.collection.IndexedSeqOptimized.foreach$(IndexedSeqOptimized.scala:33)
    at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:198)
    at scala.collection.TraversableLike.map(TraversableLike.scala:286)
    at scala.collection.TraversableLike.map$(TraversableLike.scala:279)
    at scala.collection.mutable.ArrayOps$ofRef.map(ArrayOps.scala:198)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.encoderFor(JavaTypeInference.scala:138)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.encoderFor(JavaTypeInference.scala:60)
    at org.apache.spark.sql.catalyst.JavaTypeInference$.encoderFor(JavaTypeInference.scala:53)
    at org.apache.spark.sql.catalyst.encoders.ExpressionEncoder$.javaBean(ExpressionEncoder.scala:62)
    at org.apache.spark.sql.Encoders$.bean(Encoders.scala:179)
    at org.apache.spark.sql.Encoders.bean(Encoders.scala)
    at org.example.Main.main(Main.java:26) 

I'm attaching the code I use to reproduce the error locally.  spark_test.zip

The issue is in the JavaTypeInference class when it tries to find the encoder for a ParameterizedType with the value Team<T>. When running JavaTypeUtils.getTypeArguments(pt).asScala.toMap, it returns the type T again, but this time as a Company object, and pt.getRawType as Team. This ends up generating a tuple of Team, Company in the typeVariables map, leading to errors when searching for TypeVariables.

My example code is this:

public class Main {
    public static void main(String[] args) {
        SparkSession spark = SparkSession.builder()
                .master("local[*]")
                .appName("Spark Test")
                .getOrCreate();

        Dataset<PersonData> df = spark.read()
                .option("header", "true")
                .option("delimiter", ",")
                .csv("src/main/resources/data/person-data.csv")
                .as(Encoders.bean(PersonData.class));

        Dataset<CompanyWrapper> companyWrapperDataset = df.map((MapFunction<PersonData, CompanyWrapper>) personData -> {
            Team<PersonData> team = new Team<>("TEAM NAME", personData);
            return new CompanyWrapper("COMPANY NAME", team);
        }, Encoders.bean(CompanyWrapper.class));

        companyWrapperDataset.show(false);

    }
} 

 

public class CompanyWrapper extends Company<PersonData> {

    public CompanyWrapper() {
    }

    public CompanyWrapper(String name, Team<PersonData> team) {
        super(name, team);
    }

} 

 

public class Company<T> {

    private String name;
    private Team<T> team;

    public Company() {
    }

    public Company(String name, Team<T> team) {
        this.name = name;
        this.team = team;
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }

    public Team<T> getTeam() {
        return team;
    }

    public void setTeam(Team<T> team) {
        this.team = team;
    }

    @Override
    public String toString() {
        return "Company{" +
                "name='" + name + '\'' +
                ", team=" + team +
                '}';
    }
}
 

public class Team<T> {

    public String name;
    public T person;

    public Team() {
    }

    public Team(String name, T person) {
        this.name = name;
        this.person = person;
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }

    public T getPerson() {
        return person;
    }

    public void setPerson(T person) {
        this.person = person;
    }

    @Override
    public String toString() {
        return "Team{" +
                "name='" + name + '\'' +
                ", person=" + person +
                '}';
    }
} 

public class PersonData implements Serializable {

    private String id;

    private String firstName;

    private String lastName;

    private String email;

    private String phone;

    public String getId() {
       return id;
    }

    public void setId(String id) {
       this.id = id;
    }

    public String getFirstName() {
       return firstName;
    }

    public void setFirstName(String firstName) {
       this.firstName = firstName;
    }

    public String getLastName() {
       return lastName;
    }

    public void setLastName(String lastName) {
       this.lastName = lastName;
    }

    public String getEmail() {
       return email;
    }

    public void setEmail(String email) {
       this.email = email;
    }

    public String getPhone() {
       return phone;
    }

    public void setPhone(String phone) {
       this.phone = phone;
    }

    @Override
    public String toString() {
       return "PersonData{" +
             "id='" + id + '\'' +
             ", firstName='" + firstName + '\'' +
             ", lastName='" + lastName + '\'' +
             ", email='" + email + '\'' +
             ", phone='" + phone + '\'' +
             '}';
    }
} 

 

Issue Links

• links to: GitHub Pull Request #48304
• links to: PR-for-issue