Implementation:Haifengl Smile DataType System

Knowledge Sources	Haifengl_Smile
Domains	Type System, Data Structures, Schema
Last Updated	2026-02-08 22:00 GMT

Overview

DataType is the core interface for Smile's type system, defining all supported data types and providing methods for type checking, conversion, inference, and promotion.

Description

DataType is a Serializable interface that represents the data types used throughout the Smile data framework. It defines an inner enum ID with 16 type identifiers (Boolean, Byte, Char, Short, Int, Long, Float, Double, Decimal, String, Date, Time, DateTime, Object, Array, Struct). The interface provides default methods for type classification (isPrimitive, isNumeric, isFloating, isIntegral, etc.), type promotion rules for arithmetic expressions, type coercion for schema inference, and static factory methods to construct DataType instances from strings, Java classes, or JDBC types.

Usage

Use DataType when defining schemas for DataFrames, when parsing data from files or databases, or when performing type inference on string data. It is the foundation of Smile's type system and is used by StructField, StructType, and DataTypes throughout the framework.

Code Reference

Source Location

• Repository: Haifengl_Smile
• File: base/src/main/java/smile/data/type/DataType.java
• Lines: 1-430

Signature

public interface DataType extends Serializable {
    // Inner enum
    enum ID { Boolean, Byte, Char, Short, Int, Long, Float, Double,
              Decimal, String, Date, Time, DateTime, Object, Array, Struct }

    // Abstract methods
    ID id();
    String name();
    Object valueOf(String s);

    // Default type-check methods
    default String toString(Object o);
    default boolean isNullable();
    default boolean isPrimitive();
    default boolean isFloating();
    default boolean isIntegral();
    default boolean isNumeric();
    default boolean isBoolean();
    default boolean isChar();
    default boolean isByte();
    default boolean isShort();
    default boolean isInt();
    default boolean isLong();
    default boolean isDecimal();
    default boolean isFloat();
    default boolean isDouble();
    default boolean isString();
    default boolean isObject();

    // Static utility methods
    static DataType prompt(DataType a, DataType b);
    static DataType coerce(DataType a, DataType b);
    static DataType infer(String s);
    static DataType of(String s) throws ClassNotFoundException;
    static DataType of(Class<?> clazz);
    static DataType of(java.sql.JDBCType type, boolean nullable, String dbms);
}

Import

import smile.data.type.DataType;

I/O Contract

Inputs

Name	Type	Required	Description
s	String	Yes (for valueOf/infer/of)	String representation of a value or type name.
clazz	Class<?>	Yes (for of(Class))	A Java class to map to a Smile DataType.
type	JDBCType	Yes (for of(JDBCType))	A JDBC type to convert.
nullable	boolean	Yes (for of(JDBCType))	Whether the column allows null values.
dbms	String	Yes (for of(JDBCType))	The database product name (e.g., "SQLite").

Outputs

Name	Type	Description
id()	DataType.ID	The type identifier enum value.
name()	String	The type name for external catalogs.
valueOf(s)	Object	The parsed value from a string representation.
of(s)	DataType	A DataType instance from its string representation.
infer(s)	DataType	The inferred DataType from a string value (e.g., "123" yields IntType).
prompt(a, b)	DataType	The promoted type for arithmetic between types a and b.
coerce(a, b)	DataType	The common type for schema inference between types a and b.

Usage Examples

Basic Usage

import smile.data.type.DataType;
import smile.data.type.DataTypes;

// Get a DataType from a string name
DataType intType = DataType.of("int");
DataType doubleType = DataType.of("double");

// Check type properties
boolean isNum = intType.isNumeric();    // true
boolean isPrim = intType.isPrimitive(); // true
boolean isNull = intType.isNullable();  // false

// Type promotion for expressions
DataType promoted = DataType.prompt(intType, doubleType); // DoubleType

// Infer type from string values
DataType inferred1 = DataType.infer("42");        // IntType
DataType inferred2 = DataType.infer("3.14");      // DoubleType
DataType inferred3 = DataType.infer("2024-01-15"); // DateType
DataType inferred4 = DataType.infer("hello");     // StringType

From Java Class

import smile.data.type.DataType;

// Map Java classes to Smile DataTypes
DataType dt1 = DataType.of(int.class);      // IntType
DataType dt2 = DataType.of(Double.class);   // NullableDoubleType
DataType dt3 = DataType.of(String.class);   // StringType

// Parse a value from string
Object val = DataTypes.IntType.valueOf("42"); // returns Integer 42

Related Pages

• Implementation:Haifengl_Smile_StructType