pull-level

pull-stream interface to levelup

Example - reading

read items in database.

1var pl = require('pull-level')
2var pull = require('pull-stream')
3
4var db = require('levelup')('/tmp/pull-level-example')
5
6pull(pl.read(db), pull.collect(console.log))

read items in database, plus realtime changes

1pull(
2  pl.read(db, {live: true}),
3  //log data as it comes,
4  //because tail will keep the connection open
5  //so we'll never see the end otherwise.
6  pull.through(console.log),
7  //note, pull-streams will not drain unless something is
8  //pulling the data through, so we have to add drain
9  //even though the data we want is coming from pull.through()
10  pull.drain()
11)

If you just want the realtime inserts, use live

1pull(
2  pl.live(db, {live: true}),
3  pull.through(console.log),
4  pull.drain()
5)

Example - writing

To write, pipe batch changes into write

1pull(
2  pull.values([
3    {key: 0, value: 'zero', type: 'put'},
4    {key: 1, value: 'one',  type: 'put'},
5    {key: 2, value: 'two',  type: 'put'},
6  ]),
7  pl.write(db)
8)

If you are lazy/busy, you can leave off type. In that case, if value is non-null, the change is considered a put else, a del.

1pull(
2  pull.values([
3    {key: 0, value: 'zero'},
4    {key: 1, value: 'one'},
5    {key: 2, value: 'two'},
6  ]), 
7  pl.write(db)
8)

Example - indexes!

With pull-level it's easy to create indexes. just save a pointer to the key.

like this:

1pull(
2  pull.values([
3    {key: key, value: VALUE, type: 'put'},
4    {key: '~INDEX~' + VALUE.prop, value: key,  type: 'put'},
5  ]),
6  pl.write(db)
7)

then, when you want to do a read, use asyncMap

1pull(
2  pl.read(db, {min: '~INDEX~', max: '~INDEX~~'})
3  pull.asyncMap(function (e, cb) {
4    db.get(e.value, function (value) {
5      cb(null, {key: e.value, value: value})
6    })
7  }),
8  pull.collect(console.log)
9)

Example realtime aggregation

We want to keep a realtime count of everything in the database. When ever something is inserted, we increment. But, we need to check the records that are currently in the database.

Since it takes some time to scan the database, we need to make sure we have done that before giving an answer. We can read it all with one stream, using {sync: true} to be notified of when we have read out all the old records.

First all the old records are read from the non-live stream, then you get one {sync: true} element, then all the new item.

1var sum = 0, ready = false, waiting = []
2
3//call get count to know s
4function getSum (cb) {
5  if(!ready) waiting.push(cb)
6  else cb(null, sum)
7}
8
9pull(
10  pl.read(db, {sync: true}),
11  pull.drain(function (op) {
12    if(op.sync) {
13      //if we see a data element with this it means
14      ready = true
15      while(waiting.length) waiting.shift()(null, count)
16    }
17    //increment our counter!
18    if(Number.isFinite(+op.value.amount)) //filter out non numbers & NaN.
19      sum += op.value.amount
20  })
21)
22

License

MIT