Package org.apache.tika.inference

Class VectorSerializer

java.lang.Object

org.apache.tika.inference.VectorSerializer

public final class VectorSerializer
extends Object

Serializes and deserializes float vectors as base64-encoded big-endian float32 byte arrays.

Big-endian matches the format expected by Elasticsearch's dense_vector field type, which accepts either a JSON float array or a base64-encoded binary string in big-endian float32 order. See the Elasticsearch dense_vector mapping documentation for details.

Method Summary

Modifier and Type	Method	Description
static float[]	decode(String base64)	Decode a base64 string back to a float array (big-endian float32).
static String	encode(float[] vector)	Encode a float array as a base64 string (big-endian float32).

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

encode

public static String encode(float[] vector)

Encode a float array as a base64 string (big-endian float32).

decode

public static float[] decode(String base64)

Decode a base64 string back to a float array (big-endian float32).