@qgustavor/las-reader

Read LAS lidar point clouds in Node.js and the browser.

npm install @qgustavor/las-reader

Node.js 24 or newer. ESM only.

Quick start

import { openFile } from '@qgustavor/las-reader/node'

const reader = await openFile('cloud.las')

console.log(reader.header.versionString, reader.pointCount)

for await (const point of reader) {
  console.log(point.x, point.y, point.z, point.classification)
}

await reader.close()

In a browser, open a file the user picked:

import { openBlob } from '@qgustavor/las-reader/browser'

const reader = await openBlob(fileInput.files[0])

Finding points

import { pointsNear, matching } from '@qgustavor/las-reader'

for await (const point of pointsNear(
  reader,
  { x: 2230900, y: 252200, radius: 25 },
  { where: matching({ classification: [2, 9] }) }
)) {
  console.log(point.x, point.y, point.z, point.distance)
}

Queries reject points on the stored integers before decoding anything. On a file you will query more than once, build an index first so later queries skip most of it — see Working with large files.

Documentation

| Getting started | Opening files, reading points, what a point looks like | | --- | --- | | In the browser | Local files, Web Workers, HTTP | | In Node.js | Paths, file handles, streams | | Working with large files | Block indexes, caching, memory | | Finding points | Boxes, radii, attribute predicates, columns | | Coordinate systems | What the file declares, and reprojecting with proj4 | | Byte sources | The I/O interface and writing your own | | Compressed files | Reading .laz, chunk tables, sharing a WASM module | | API reference | Every export | | Migrating from 1.x | What changed and why |

Compressed files

LASzip (.laz) files open the same way .las files do:

const reader = await openFile('cloud.laz')

The openers read the header and pick the right reader, so callers do not have to know which they have. Decompression uses laz-perf in WebAssembly, loaded only when a compressed file is actually opened, so a bundle that reads only .las never pulls it in.

Chunks are decompressed as they are needed, so readPoint(4_000_000) costs one chunk rather than the whole file — see Compressed files.

License

Apache-2.0