Map Types

Map types store key-value pairs within a single column. Support varies by database.

Database Support

Database	Mechanism	Status
PostgreSQL	hstore extension	Mapped to Map<String, String>
DuckDB	MAP(key, value)	Full support with typed Map<K, V>
MariaDB	-	No map support
Oracle	-	No map support
SQL Server	-	No map support
DB2	-	No map support

PostgreSQL hstore

PostgreSQL's hstore extension provides key-value storage with string keys and values:

CREATE EXTENSION IF NOT EXISTS hstore;

CREATE TABLE product (
  id SERIAL PRIMARY KEY,
  attributes hstore
);

INSERT INTO product (attributes)
VALUES ('color => red, size => large');

Typr maps hstore to Map<String, String>:

// Java
Map<String, String> attributes = product.attributes();

// Kotlin
val attributes: Map<String, String>? = product.attributes

// Scala
val attributes: Option[Map[String, String]] = product.attributes

DuckDB MAP

DuckDB supports typed MAP columns with arbitrary key and value types:

CREATE TABLE sensor_readings (
  sensor_id UUID PRIMARY KEY,
  hourly_temps MAP(INTEGER, DOUBLE),     -- hour -> temperature
  daily_counts MAP(DATE, INTEGER)         -- date -> count
);

Typr generates typed Map<K, V> with full read/write support:

// Java - typed map access
Map<Integer, Double> temps = reading.hourlyTemps();
Double noonTemp = temps.get(12);

Map<LocalDate, Integer> counts = reading.dailyCounts();

// Kotlin
val temps: Map<Int, Double> = reading.hourlyTemps
val noonTemp = temps[12]

// Scala
val temps: Map[Int, Double] = reading.hourlyTemps
val noonTemp = temps.get(12)

DuckDB MAPs support any combination of key and value types that DuckDB supports.

Databases Without Map Support

For databases without native map types, alternatives include:

• JSON columns - Store maps as JSON objects
• Separate tables - Use a key-value table with foreign keys
• Delimited strings - Simple but limited (no type safety)