leveled

node-leveled

A js-style LevelDB binding for node.

The datatype is plain strings, support for buffers and other formats may be added.

Usage

var db = require('leveled')('/tmp/mydb');

db.putSync('foo', 'bar')

db.get('some', function (err, val) {
  val == 'value';
})

db.find('*', function (err, res) {
  console.log(res); // { 'a' : 'foo', 'aa' : 'baz', ... }  
})

db.batch()
  .put('foo', 'bar')
  .del('baz')
  .write(function (err) {
    // success
  })

Installation

$ npm install leveled

API

leveled(path)

Instantiates a new DB at path, creating path if necessary.

leveled#put(key, val[, cb])

leveled#putSync(key, val)

Store val at key.

leveled#get(key, cb)

leveled#getSync(key)

Get the value stored at key.

leveled#del(key[, cb])

leveled#delSync(key)

Delete the value stored at key.

leveled#range(from, to, cb)

Find all entries whose keys are in the give range.

• from can be from / [from or (from
• to can be to / to] or to)

db.range('[1337', '2337)', function (err, res) {
  console.log(res) // { 1337 : 'foo', ..., 2336 : 'foo' }  
})

leveled#find(glob, cb)

Find values.

db.find('ab*', function (err, res) {
  console.log(res) // { 'aba' : 'foo', 'abzzz' : 'bar' }
})

At the moment glob-style matching is not fully implemented, what works is

• abc*
• *

In the future it will work like redis's KEYS command:

• ab? matches abc, not abcc
• ab[1-3] matches ab1, ab2, ab3
• ab[1-3]* etc.

leveled#batch()

Creates a new batch that queues up operations until its write method is invoked.

batch#put(key, val)

batch#del(key)

batch#write(cb)

batch#writeSync()

Apply the batch's operations to the DB.

Benchmark

On my mb pro:

$ node bench/bench.js

  benchmarking with 120,000 records, 24 chars each

          put :   128,479 w/s in    934ms
      putSync :   372,670 w/s in    322ms

        batch :   612,244 w/s in    196ms
    batchSync :   641,711 w/s in    187ms

          get :    58,881 r/s in  2,038ms
      getSync :   560,747 r/s in    214ms

     iterator :   220,588 r/s in    544ms

put oparations don't force a sync to disk, hence the get looking so slow. Iterators are faster for common tasks where you need to get many values anyways.

TODO

• evaluate leveldb::Slice for storage
• evaluate buffers as data type
• evaluate storing native js objects
• evaluate msgpack

License

(MIT)