node package manager



join leveldb documents based on common nested values

build status


Given this dataset with commit and output rows:

  { "type": "commit", "id": 444, "hash": "6eba8e6a2927a5d8b748d422ad7e64b977ab4f94", "time": 1374904364802 },
  { "type": "output", "job": 444, "data": "beep ", "time": 1374904288007 },
  { "type": "output", "job": 444, "data": "boop.", "time": 1374904289854 },
  { "type": "commit", "id": 555, "hash": "5c825a710662cab0b8abb37132cae19d0dcf00cb", "time": 1374904278950 },
  { "type": "output", "job": 555, "data": "hello ", "time": 1374904366509 },
  { "type": "output", "job": 555, "data": "world!", "time": 1374904367169 }

we can populate the database with db.batch() and then join each output's job property with a matching commit id property. This is like a foreign key in an SQL database, except that it can be arbitrarily nested.

var sub = require('level-sublevel');
var level = require('level');
var through = require('through');
var shasum = require('shasum');
var db = sub(level('/tmp/testdb', { valueEncoding: 'json' }));
var join = require('level-join')(db);
db.batch(require('./data.json').map(function (row) {
    var key = shasum(row);
    return { type: 'put', key: key, value: row };
join.add('commit', [ 'id' ], [ 'type', 'commit' ]);
join.add('output', [ 'job' ], [ 'type', 'output' ]);
join.pipe(through(function (pair) {

For each pairing of commit and output objects, the commit hash and output data get printed:

5c825a710662cab0b8abb37132cae19d0dcf00cb hello 
6eba8e6a2927a5d8b748d422ad7e64b977ab4f94 boop.
5c825a710662cab0b8abb37132cae19d0dcf00cb world!
6eba8e6a2927a5d8b748d422ad7e64b977ab4f94 beep 

Note that the results are not grouped or sorted. A grouping and sorting module could be of use here.


var levelJoin = require('level-join')

var join = levelJoin(db)

Create a new join instance from a [sublevel](]-capable leveldb handle db.

join is an object-mode readable stream that outputs pair objects with keys for each of the names that have been added with join.add() and values of the matching documents.

join.add(name, joinPath, filterPath)

Add name to the join using the value at the pathway array joinPath and filtering results by the filterPath array that will be passed through directly to level-search.


With npm do:

npm install level-join