Implementation:Lance format Lance Java ProjectOp

Knowledge Sources	Lance
Domains	Java_SDK, Dataset_Management
Last Updated	2026-02-08 19:33 GMT

Overview

Description

The Project class is an immutable operation that changes the schema of a Lance dataset without modifying the underlying data. It extends SchemaOperation (which implements Operation) and carries only an Arrow Schema as its payload.

Project modifies column positions but does not change field IDs. When columns are removed via a Project operation, the physical data files are not immediately deleted; the data is removed after a subsequent compaction operation. This makes Project a lightweight metadata-only operation.

Usage

Use Project for schema-only transformations such as removing columns (logically), reordering columns, or renaming columns. Since it does not touch data files, it executes quickly regardless of dataset size. Physical cleanup of removed column data occurs during compaction.

Code Reference

Source Location

java/src/main/java/org/lance/operation/Project.java

Signature

public class Project extends SchemaOperation {
    public static Builder builder();
    public Schema schema();       // inherited from SchemaOperation
    public long exportSchema(BufferAllocator allocator); // inherited
    public String name();         // returns "Project"
}

Import

import org.lance.operation.Project;

I/O Contract

Inputs

Parameter	Type	Required	Description
schema	org.apache.arrow.vector.types.pojo.Schema	Yes	The new schema to project the dataset to

Outputs

Return	Type	Description
schema()	Schema	The projected schema
exportSchema(allocator)	long	Memory address of the exported Arrow C schema for JNI
name()	String	Returns "Project" for JNI dispatch

Usage Examples

// Remove a column by projecting to a schema without it
Schema projectedSchema = new Schema(List.of(
    Field.nullable("id", new ArrowType.Int(64, true)),
    Field.nullable("name", new ArrowType.Utf8())
    // "deprecated_column" is excluded from the projection
));

Project projectOp = Project.builder()
    .schema(projectedSchema)
    .build();

String opName = projectOp.name(); // "Project"

Related Pages

• Lance_format_Lance_Java_MergeOp -- Adds columns with data via schema evolution
• Lance_format_Lance_Java_OverwriteOp -- Overwrites the entire dataset including schema
• Lance_format_Lance_Java_ColumnAlteration -- Alters column properties such as rename, nullable, and type cast