Tiny and fast async generator to read streams line by line

Node 10+ is required. This package provides very simple but feature-reach functionality to consume readable streams or files line by line. It's also quite fast (up to 30% faster than readline) and has no dependencies.

Features:

• reading of files (by name)
• reading of readable streams
• generator mode via readlines
• readable stream mode via LineStream
• customizable separator (default is \n)
• raw bytes mode if encoding option is null
• configurable lookahead buffer size

Documentation

Have a look at the index.js, it's really tiny. Or check the usage example below.

Usage examples:

const { readlines, LineStream } = require('readlines-ng');

// As a generator
for await (const line of readlines(filename, { encoding: 'utf8' })) {
  console.log(line);
}

// As a readable stream
const reader = new LineStream(filename, {
  chunkSize: 256*1024,
  encoding: null,
});
for await (const line of reader) {
  console.log(line);
}

// Custom separator
const { PassThrough } = require('stream');
const datasource = new PassThrough();
datasource.write('qux42');
datasource.write('foobar42');
datasource.end();

const reader = new LineStream(filename, { sep: '42' });
for await (const line of reader) {
  console.log(line);
}

Tests

npm run test

Benchmark

Trivial wc -l impl using this package exposes the performance better than the standard readline module.

npm run bench </path/to/huge/file>  # Node 11+ is required

# File size ~2.8 GB
> Benchmark: lines counter (readline module)
> 5974719 lines found
> Done in 5.757290349 sec

> Benchmark: lines counter (async generator)
> 5974719 lines found
> Done in 4.707053160 sec

> Benchmark: lines counter (async generator UTF-8)
> 5974719 lines found
> Done in 3.937784544 sec

> Benchmark: lines counter (readable stream)
> 5974719 lines found
> Done in 4.984460949 sec

> Benchmark: lines counter (readable stream UTF-8)
> 5974719 lines found
> Done in 4.354594750 sec