b64Vecs

Utilities for encoding and decoding points using base64 and Float16 encoding. Provides functions for converting between VecModel arrays and compact base64 strings, as well as individual point encoding/decoding operations.

class b64Vecs {}

Methods

Get the first point from a delta-encoded base64 string. The first point is stored as Float32 for full precision.

static decodeFirstPoint(b64Points: string): null | VecModel

Parameters

string

Returns

null | VecModel

The first point as a VecModel, or null if the string is too short

Get the last point from a delta-encoded base64 string. Requires decoding all points to accumulate deltas.

static decodeLastPoint(b64Points: string): null | VecModel

Parameters

string

Returns

null | VecModel

The last point as a VecModel, or null if the string is too short

Decode a delta-encoded base64 string back to an array of absolute VecModels. The first point is stored as Float32 (high precision), subsequent points are Float16 deltas that are accumulated to reconstruct absolute positions.

static decodePoints(base64: string): VecModel[]

Parameters

string

Returns

VecModel[]

An array of VecModel objects with absolute coordinates

Encode an array of VecModels using delta encoding for improved precision. The first point is stored as Float32 (high precision for absolute position), subsequent points are stored as Float16 deltas from the previous point. This provides full precision for the starting position and excellent precision for deltas between consecutive points (which are typically small values).

Format: - First point: 3 Float32 values = 12 bytes = 16 base64 chars - Delta points: 3 Float16 values each = 6 bytes = 8 base64 chars each

static encodePoints(points: VecModel[]): string

Parameters

VecModel[]

Returns

string

A base64-encoded string containing delta-encoded points