vcti-datanode 1.0.1

Lightweight data-plus-attributes node containers, with an optional lazily-loaded variant for out-of-core data.

Data Node

Lightweight data-plus-attributes node containers, with an optional lazily-loaded variant for out-of-core data.

Overview

vcti-datanode provides two small, tree-agnostic value types:

• DataNode — pairs data (an np.ndarray, any array-like, or an arbitrary object) with an attributes metadata dict. Value equality and readable repr/str. That's it.
• LazyDataNode — a DataNode whose data is fetched on demand through a loader callback (load()) and can be freed (release()) and reloaded later. Attributes stay resident; only the data comes and goes. Built for large datasets (e.g. CAE models) where holding everything in memory at once is impractical.

They are commonly used as node payloads in a tree, but depend on nothing but numpy and can be used anywhere a "data + metadata" unit is useful.

Installation

pip install vcti-datanode

In requirements.txt

vcti-datanode>=1.0.1

In pyproject.toml dependencies

dependencies = [
    "vcti-datanode>=1.0.1",
]

---

Quick Start

import numpy as np
from vcti.datanode import DataNode, LazyDataNode

# Eager: data held directly.
node = DataNode(data=np.array([1.0, 2.0, 3.0]), attributes={"units": "mm"})

# Lazy: data fetched on demand, freed when not needed.
lazy = LazyDataNode(loader=lambda: np.load("displacement.npy"),
                    attributes={"name": "displacement"})
lazy.is_loaded          # False
arr = lazy.load()       # loader runs once; cached thereafter
lazy.release()          # frees the data; attributes remain

---

Dependencies

• numpy>=1.26 — used for array-aware value equality and repr.