Skip to content
JavaScript implementation of the UnixFs importer used by IPFS
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
.npmignore data exporting Apr 19, 2016
LICENSE feat: switch to async await (#24) May 17, 2019
package.json chore: update contributors Aug 5, 2019


standard-readme compliant Build Status Codecov Dependency Status js-standard-style

JavaScript implementation of the layout and chunking mechanisms used by IPFS to handle Files

Lead Maintainer

Alex Potsides

Table of Contents


> npm install ipfs-unixfs-importer



Let's create a little directory to import:

> cd /tmp
> mkdir foo
> echo 'hello' > foo/bar
> echo 'world' > foo/quux

And write the importing logic:

const importer = require('ipfs-unixfs-importer')

// Import path /tmp/foo/bar
const source = [{
  path: '/tmp/foo/bar',
  content: fs.createReadStream(file)
}, {
  path: '/tmp/foo/quxx',
  content: fs.createReadStream(file2)

// You need to create and pass an ipld-resolve instance
for await (const entry of importer(source, ipld, options)) {

When run, metadata about DAGNodes in the created tree is printed until the root:

  cid: CID, // see
  path: 'tmp/foo/bar',
  unixfs: UnixFS // see
  cid: CID, // see
  path: 'tmp/foo/quxx',
  unixfs: UnixFS // see
  cid: CID, // see
  path: 'tmp/foo',
  unixfs: UnixFS // see
  cid: CID, // see
  path: 'tmp',
  unixfs: UnixFS // see


const importer = require('ipfs-unixfs-importer')

const import = importer(source, ipld [, options])

The import function returns an async iterator takes a source async iterator that yields objects of the form:

  path: 'a name',
  content: (Buffer or iterator emitting Buffers)

import will output file info objects as files get stored in IPFS. When stats on a node are emitted they are guaranteed to have been written.

ipld is an instance of the IPLD Resolver or the js-ipfs dag api

The input's file paths and directory structure will be preserved in the dag-pb created nodes.

options is an JavaScript option that might include the following keys:

  • wrap (boolean, defaults to false): if true, a wrapping node will be created
  • shardSplitThreshold (positive integer, defaults to 1000): the number of directory entries above which we decide to use a sharding directory builder (instead of the default flat one)
  • chunker (string, defaults to "fixed"): the chunking strategy. Supports:
    • fixed
    • rabin
  • chunkerOptions (object, optional): the options for the chunker. Defaults to an object with the following properties:
    • avgChunkSize (positive integer, defaults to 262144): the average chunk size (rabin chunker only)
    • minChunkSize (positive integer): the minimum chunk size (rabin chunker only)
    • maxChunkSize (positive integer, defaults to 262144): the maximum chunk size
  • strategy (string, defaults to "balanced"): the DAG builder strategy name. Supports:
    • flat: flat list of chunks
    • balanced: builds a balanced tree
    • trickle: builds a trickle tree
  • maxChildrenPerNode (positive integer, defaults to 174): the maximum children per node for the balanced and trickle DAG builder strategies
  • layerRepeat (positive integer, defaults to 4): (only applicable to the trickle DAG builder strategy). The maximum repetition of parent nodes for each layer of the tree.
  • reduceSingleLeafToSelf (boolean, defaults to true): optimization for, when reducing a set of nodes with one node, reduce it to that node.
  • dirBuilder (object): the options for the directory builder
    • hamt (object): the options for the HAMT sharded directory builder
      • bits (positive integer, defaults to 8): the number of bits at each bucket of the HAMT
  • progress (function): a function that will be called with the byte length of chunks as a file is added to ipfs.
  • onlyHash (boolean, defaults to false): Only chunk and hash - do not write to disk
  • hashAlg (string): multihash hashing algorithm to use
  • cidVersion (integer, default 0): the CID version to use when storing the data (storage keys are based on the CID, including it's version)
  • rawLeaves (boolean, defaults to false): When a file would span multiple DAGNodes, if this is true the leaf nodes will not be wrapped in UnixFS protobufs and will instead contain the raw file bytes
  • leafType (string, defaults to 'file') what type of UnixFS node leaves should be - can be 'file' or 'raw' (ignored when rawLeaves is true)


Feel free to join in. All welcome. Open an issue!

This repository falls under the IPFS Code of Conduct.



You can’t perform that action at this time.