multifeed-index

multifeed-index

Build an index over a set of hypercores.

Traverses a set of hypercores as a multifeed and calls a user indexing function to build an index.

Purpose

A multifeed is a set of append-only logs (feeds), with the property that only the feed's author can write new entries to it.

One type of document that you might write to such a feed are key -> value pairs. Maybe documents looks like this:

{
  id: '23482934',
  name: 'quinn',
  gender: null
}

So some key, '23482934', can map to this document. How could you make the lookup from '23482934' to the aforementioned document fast? What if there are thousands of such entries?

You'd probably want to build some kind of index. One that iterates over every entry in every feed, and also listens for new entries that get added. Then you could create an efficient data structure (say, maybe with level) that can map each key to a value quickly.

Good news: this is essentially what multifeed-index does!

It does this by taking a multifeed you give it, along with three functions that the user provides:

• storeState(state, cb): multifeed-index will give you a Buffer object, state, to store somewhere of your choosing. This could be in memory, to a database, to a JSON file, to browser storage, or whatever makes sense for your program. The module doesn't want to tie you down to a specific storage method.
• fetchState(cb): Like storeState, except this function will be called by multifeed-index when it needs to retrieve the state. Your job is to call the callback cb with the same state Buffer that was given to storeState earlier.
• clearIndex(cb): This only gets called when you change the version of the multifeed-index (see below). This function should delete the entire index and whatever state it stored earlier, so that the new version of the index can be regenerated from scratch.

Example

Let's build a key-value index using this and unordered-materialized-kv:

var multifeed = require('multifeed')
var indexer = require('multifeed-index')
var umkv = require('unordered-materialized-kv')
var ram = require('random-access-memory')
var memdb = require('memdb')

var multi = multifeed(ram, { valueEncoding: 'json' })

var kv = umkv(memdb())

var kvView = indexer({
  version: 1,  // setting a different number will cause the index to be purged and rebuilt
  log: multi,
  maxBatch: 5,
  batch: function (nodes, next) {
    var ops = nodes.map(function (node) {
      return {
        id: node.key.toString('hex') + '@' + node.seq,
        key: node.value.key,
        links: node.value.links
      }
    })
    kv.batch(ops, next)
  }
})

function append (w, data, cb) {
  w.append(data, function (err) {
    if (err) return cb(err)
    var id = w.key.toString('hex') + '@' + (w.length - 1)
    cb(null, id)
  })
}

multi.writer(function (_, w) {
  append(w, {
    key: 'foo',
    value: 'bax',
    links: []
  }, function (_, id1) {
    console.log('id-1', id1)
    append(w, {
      key: 'foo',
      value: 'bax',
      links: [id1]
    }, function (_, id2) {
      console.log('id-2', id2)
      kvView.ready(function () {
        kv.get('foo', function (_, res) {
          console.log(res)
        })
      })
    })
  })
})

outputs

id-1 9736a3ff7ae522ca80b7612fed5aefe8cfb40e0a43199174e47d78703abaa22f@0
id-2 9736a3ff7ae522ca80b7612fed5aefe8cfb40e0a43199174e47d78703abaa22f@1
[
  '9736a3ff7ae522ca80b7612fed5aefe8cfb40e0a43199174e47d78703abaa22f@1'
]

API

var Index = require('multifeed-index')

var index = Index(opts)

Required opts include:

• log: a multifeed instance
• batch: a mapping function, to be called on 1+ nodes at a time in the hypercores of log.
• storeState: Function of the form function (state, cb). Called by the indexer when there is a new indexing state to save. The user can store this Buffer object whereever/however they'd like.
• fetchState: Function of the form function (cb). Called by the indexer to seed the indexing process when the indexer is created. Expects cb to be called with (err, state), where state is a Buffer that was previously given to opts.storeState.

The batch function expects params (nodes, next). next should be called when you are done processing. Each node of nodes is of the form

{
  key: 'hex-string-of-hypercore-public-key',
  seq: 14,
  value: <whatever value is in the hypercore at this feed + sequence number>
}

Optional opts include:

• version: the version of the index. If increased, any indexes built with an earlier version will be purged and rebuilt. This is useful for when you change the data format of the index and want all peers to rebuild to use this format. Defaults to 1.
• maxBatch: maximum batch size of nodes to process in one batch call. Defaults to 50.
• clearIndex: Function of the form function (cb). Called by the indexer when a new version for the index has been passed in (via opts.version) and the index needs to be cleared & regenerated.

index.ready(cb)

Registers the callback cb() to fire when the indexes have "caught up" to the latest known change. The cb() function fires exactly once. You may call index.ready() multiple times with different functions.

index.pause([cb])

Pauses the indexing process. Whatever batches of entries are currently being processed will finish first. If a callback cb is given, it will be called once pending entries are processed and the indexer is fully paused.

index.resume()

Synchronously restarts a paused indexer.

var state = index.getState()

Returns the current state of the indexer.

state is an object that looks like

{
  state: 'indexing',                // one of ['idle', 'indexing', 'paused', 'error']
  context: {
    totalBlocks: 50,                // total # of blocks known of across all feeds
    indexedBlocks: 18,              // total # of blocks indexed so far
    prevIndexedBlocks: 0,           // total # of blocks indexed as of the previous indexing run
    indexStartTime: 1583184761581,  // ms since epoch when the last indexing run started
    error: null                     // error state (indexing is unusable if this is set)
  }
}

index.on('indexed', function (nodes) {})

Event emitted when entries have finished being indexed.

index.on('error', function (err) {})

Event emitted when an error within multifeed-index has occurred. This is very important to listen on, lest things suddenly seem to break and it's not immediately clear why.

index.on('state-update', function (state) {})

Event emitted when the internal state of the indexer changes. It has the same form as the state object returned by index.getState() above.

Install

$ npm install multifeed-index

See Also

• hyperlog-index
• hyperdb-index

License

ISC