The files API enables users to use the File System abstraction of IPFS. There are two Files API, one at the top level, the original add, cat, get and ls, and another behind the files, also known as MFS. We are currently going through a revamping process of these APIs to make them more user-friendly.

The regular, top-level API for add, cat, get and ls Files on IPFS

• add
• cat
• get
• ls

The Files API, aka MFS (Mutable File System)

• files.chmod
• files.cp
• files.flush
• files.ls
• files.mkdir
• files.mv
• files.read
• files.rm
• files.stat
• files.touch
• files.write

Explore the Mutable File System through interactive coding challenges in our ProtoSchool tutorial.

Import files and data into IPFS.

Where data may be:

• Bytes (alias for Buffer|ArrayBuffer|TypedArray) [single file]
• Bloby (alias for: Blob|File) [single file]
• string [single file]
• FileObject (see below for definition) [single file]
• Iterable<number> [single file]
• Iterable<Bytes> [single file]
• Iterable<Bloby> [multiple files]
• Iterable<string> [multiple files]
• Iterable<FileObject> [multiple files]
• AsyncIterable<Bytes> [single file]
• AsyncIterable<Bloby> [multiple files]
• AsyncIterable<String> [multiple files]
• AsyncIterable<FileObject> [multiple files]

FileObject is a plain JS object of the following form:

{
  // The path you want to the file to be accessible at from the root CID _after_ it has been added
  path?: string
  // The contents of the file (see below for definition)
  content?: FileContent
  // File mode to store the entry with (see https://en.wikipedia.org/wiki/File_system_permissions#Numeric_notation)
  mode?: number | string
  // The modification time of the entry (see below for definition)
  mtime?: UnixTime
}

If no path is specified, then the item will be added to the root level and will be given a name according to it's CID.

If no content is passed, then the item is treated as an empty directory.

One of path or content must be passed.

FileContent is one of the following types:

Bytes | Bloby | string | Iterable<number> | Iterable<Bytes> | AsyncIterable<Bytes>

UnixTime is one of the following types:

Date | { secs: number, nsecs?: number } | number[]

As an object, secs is the number of seconds since (positive) or before (negative) the Unix Epoch began and nsecs is the number of nanoseconds since the last full second.

As an array of numbers, it must have two elements, as per the output of process.hrtime().

options is an optional object argument that might include the following keys:

• chunker (string, default size-262144): chunking algorithm used to build ipfs DAGs. Available formats:
  • size-{size}
  • rabin
  • rabin-{avg}
  • rabin-{min}-{avg}-{max}
• cidVersion (integer, default 0): the CID version to use when storing the data (storage keys are based on the CID, including its version).
• enableShardingExperiment: allows to create directories with an unlimited number of entries currently size of unixfs directories is limited by the maximum block size. Note that this is an experimental feature.
• hashAlg (string, default sha2-256): multihash hashing algorithm to use. The list of all possible values.
• onlyHash (boolean, default false): doesn't actually add the file to IPFS, but rather calculates its hash.
• pin (boolean, default true): pin this object when adding.
• progress (function): a function that will be called with the byte length of chunks as a file is added to ipfs.
• rawLeaves (boolean, default false): if true, DAG leaves will contain raw file data and not be wrapped in a protobuf.
• shardSplitThreshold (integer, default 1000): specifies the maximum size of unsharded directory that can be generated.
• trickle (boolean, default false): if true will use the trickle DAG format for DAG generation. Trickle definition from go-ipfs documentation.
• wrapWithDirectory (boolean, default false): adds a wrapping node around the content.

Returns

Each yielded object is of the form:

{
  path: '/tmp/myfile.txt',
  cid: CID('QmHash'),
  mode: Number,
  mtime: { secs: Number, nsecs: Number },
  size: 123
}

Example:

const files = [{
  path: '/tmp/myfile.txt',
  content: 'ABC'
}]

for await (const result of ipfs.add(content)) {
  console.log(result)
}

/*
Prints out objects like:

{
  "path": "tmp",
  "cid": CID("QmWXdjNC362aPDtwHPUE9o2VMqPeNeCQuTBTv1NsKtwypg"),
  "mode": 493,
  "mtime": { secs: Number, nsecs: Number },
  "size": 67
}

{
  "path": "/tmp/myfile.txt",
  "cid": CID("QmNz1UBzpdd4HfZ3qir3aPiRdX5a93XwTuDNyXRc6PKhWW"),
  "mode": 420,
  "mtime": { secs: Number, nsecs: Number },
  "size": 11
}
*/

Now ipfs.io/ipfs/Qm...WW returns the "ABC" string.

Both js-ipfs and js-ipfs-http-client export a utility to make importing files from the file system easier (Note: it not available in the browser).

const IPFS = require('ipfs')
const { globSource } = IPFS

const ipfs = await IPFS.create()

for await (const file of ipfs.add(globSource('./docs', { recursive: true }))) {
  console.log(file)
}

/*
{
  path: 'docs/assets/anchor.js',
  cid: CID('QmVHxRocoWgUChLEvfEyDuuD6qJ4PhdDL2dTLcpUy3dSC2'),
  size: 15347
}
{
  path: 'docs/assets/bass-addons.css',
  hash: CID('QmPiLWKd6yseMWDTgHegb8T7wVS7zWGYgyvfj7dGNt2viQ'),
  size: 232
}
...
*/

Both js-ipfs and js-ipfs-http-client export a utility to make importing a file from a URL easier.

const IPFS = require('ipfs')
const { globSource } = IPFS

const ipfs = await IPFS.create()

for await (const file of ipfs.add(urlSource('https://ipfs.io/images/ipfs-logo.svg'))) {
  console.log(file)
}

/*
{
  path: 'ipfs-logo.svg',
  cid: CID('QmTqZhR6f7jzdhLgPArDPnsbZpvvgxzCZycXK7ywkLxSyU'),
  size: 3243
}
*/

A great source of examples can be found in the tests for this API.

Returns a file addressed by a valid IPFS Path.

ipfsPath can be of type:

• CID of type:
  • string - the base encoded version of the CID
  • CID - a CID instance
  • Buffer - the raw Buffer of the CID
• string - including the ipfs handler, a CID and a path to traverse to, e.g.
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66'
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'
  • 'QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'

options is an optional object that may contain the following keys:

• offset is an optional byte offset to start the stream at
• length is an optional number of bytes to read from the stream

Returns

Example:

const chunks = []
for await (const chunk of ipfs.cat(ipfsPath)) {
  chunks.push(chunk)
}
console.log(Buffer.concat(chunks).toString())

A great source of examples can be found in the tests for this API.

Fetch a file or an entire directory tree from IPFS that is addressed by a valid IPFS Path.

ipfsPath can be of type:

• CID of type:
  • string - the base encoded version of the CID
  • CID - a CID instance
  • Buffer - the raw Buffer of the CID
• String, including the ipfs handler, a cid and a path to traverse to, e.g.
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66'
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'
  • 'QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'

Returns

Each yielded object is of the form:

{
  path: string,
  content: <AsyncIterable<BufferList>>,
  mode: number,
  mtime: { secs: number, nsecs: number }
}

Here, each path corresponds to the name of a file, and content is an async iterable with the file contents.

Example:

const BufferList = require('bl/BufferList')
const cid = 'QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF'

for await (const file of ipfs.get(cid)) {
  console.log(file.path)

  const content = new BufferList()
  for await (const chunk of file.content) {
    content.append(chunk)
  }

  console.log(content.toString())
}

A great source of examples can be found in the tests for this API.

Lists a directory from IPFS that is addressed by a valid IPFS Path.

ipfsPath can be of type:

• CID of type:
  • string - the base encoded version of the CID
  • CID - a CID instance
  • Buffer - the raw Buffer of the CID
• String, including the ipfs handler, a cid and a path to traverse to, e.g.
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66'
  • '/ipfs/QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'
  • 'QmXEmhrMpbVvTh61FNAxP9nU7ygVtyvZA8HZDUaqQCAb66/a.txt'

Returns

Each yielded object is of the form:

{
  depth: 1,
  name: 'alice.txt',
  path: 'QmVvjDy7yF7hdnqE8Hrf4MHo5ABDtb5AbX6hWbD3Y42bXP/alice.txt',
  size: 11696,
  cid: CID('QmZyUEQVuRK3XV7L9Dk26pg6RVSgaYkiSTEdnT2kZZdwoi'),
  type: 'file',
  mode: Number,
  mtime: { secs: Number, nsecs: Number }
}

Example:

const cid = 'QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF'

for await (const file of ipfs.ls(cid)) {
  console.log(file.path)
}

A great source of examples can be found in the tests for this API.

The Mutable File System (MFS) is a virtual file system on top of IPFS that exposes a Unix like API over a virtual directory. It enables users to write and read from paths without having to worry about updating the graph. It enables things like ipfs-blob-store to exist.

Change mode for files and directories

Where:

• path is the path to the entry to modify. It might be:
  • An existing MFS path to a file or a directory (e.g. /my-dir/my-file.txt)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• mode is the new file mode. It might be:
  • A string modification of the existing mode, e.g. 'a+x', 'g-w', etc
  • An integer, e.g. the returned value from parseInt('0755', 8) or 0o755
• options is an optional Object that might contain the following keys:
  • recursive is a Boolean value that indicates if mode should be applied to all sub files/directories of path (default: false)
  • hashAlg is which algorithm to use when creating CIDs for modified entries. (default: sha2-256) The list of all possible values
  • flush is a Boolean value to decide whether or not to immediately flush MFS changes to disk (default: true)
  • cidVersion: the CID version to use for any updated entries (integer, default 0)

Returns

Example:

// To give a file -rwxrwxrwx permissions
await ipfs.files.chmod('/path/to/file.txt', parseInt('0777', 8))

// Alternatively
await ipfs.files.chmod('/path/to/file.txt', '+rwx')

// You can omit the leading `0` too
await ipfs.files.chmod('/path/to/file.txt', '777')

Copy files.

Where:

• from is the path(s) of the source to copy. It might be:
  • An existing MFS path to a file or a directory (e.g. /my-dir/my-file.txt)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• to is the path of the destination to copy to
• options is an optional Object that might contain the following keys:
  • parents is a Boolean value to decide whether or not to make the parent directories if they don't exist (default: false)
  • hashAlg is which algorithm to use when creating CIDs for newly created directories. (default: sha2-256) The list of all possible values
  • flush is a Boolean value to decide whether or not to immediately flush MFS changes to disk (default: true)

If from has multiple values then to must be a directory.

If from has a single value and to exists and is a directory, from will be copied into to.

If from has a single value and to exists and is a file, from must be a file and the contents of to will be replaced with the contents of from otherwise an error will be returned.

If from is an IPFS path, and an MFS path exists with the same name, the IPFS path will be chosen.

If from is an IPFS path and the content does not exist in your node's repo, only the root node of the source file with be retrieved from the network and linked to from the destination. The remainder of the file will be retrieved on demand.

Returns

Example:

// To copy a file
await ipfs.files.cp('/src-file', '/dst-file')

// To copy a directory
await ipfs.files.cp('/src-dir', '/dst-dir')

// To copy multiple files to a directory
await ipfs.files.cp('/src-file1', '/src-file2', '/dst-dir')

Make a directory.

Where:

• path is the path to the directory to make
• options is an optional Object that might contain the following keys:
  • parents is a Boolean value to decide whether or not to make the parent directories if they don't exist (default: false)
  • hashAlg is which algorithm to use when creating CIDs for newly created directories (default: sha2-256) The list of all possible values
  • flush is a Boolean value to decide whether or not to immediately flush MFS changes to disk (default: true)
  • mode: optional UnixFS mode to create the directory with - a number or a string that will be interpreted as a base 8 number
  • mtime: A Date object, an object with { secs, nsecs } properties where secs is the number of seconds since (positive) or before (negative) the Unix Epoch began and nsecs is the number of nanoseconds since the last full second, or the output of process.hrtime()

Returns

Example:

await ipfs.files.mkdir('/my/beautiful/directory')

Get file or directory status.

Where:

• path is the path to the file or directory to stat. It might be:
  • An existing MFS path to a file or directory (e.g. /my-dir/a.txt)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• options is an optional Object that might contain the following keys:
  • hash is a Boolean value to return only the hash (default: false)
  • size is a Boolean value to return only the size (default: false)
  • withLocal is a Boolean value to compute the amount of the dag that is local, and if possible the total size (default: false)

Returns

the returned object has the following keys:

• cid a CID instance
• size is an integer with the file size in Bytes
• cumulativeSize is an integer with the size of the DAGNodes making up the file in Bytes
• type is a string that can be either directory or file
• blocks if type is directory, this is the number of files in the directory. If it is file it is the number of blocks that make up the file
• withLocality is a boolean to indicate if locality information is present
• local is a boolean to indicate if the queried dag is fully present locally
• sizeLocal is an integer indicating the cumulative size of the data present locally

Example:

const stats = await ipfs.files.stat('/file.txt')
console.log(stats)

// {
//   hash: CID('QmXmJBmnYqXVuicUfn9uDCC8kxCEEzQpsAbeq1iJvLAmVs'),
//   size: 60,
//   cumulativeSize: 118,
//   blocks: 1,
//   type: 'file'
// }

Update the mtime of a file or directory

Where:

• path is the path to the file or directory to update. It might be:
  • An existing MFS path to a file or directory (e.g. /my-dir/a.txt)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• options is an optional Object that might contain the following keys:
  • mtime Either a Date object, an object with { sec, nsecs } properties or the output of process.hrtime() (default: now)
  • hashAlg is which algorithm to use when creating CIDs for modified entries. (default: sha2-256) The list of all possible values
  • flush is a Boolean value to decide whether or not to immediately flush MFS changes to disk (default: true)
  • cidVersion: the CID version to use for any updated entries (integer, default 0)

Returns

Example:

// set the mtime to the current time
await ipfs.files.touch('/path/to/file.txt')

// set the mtime to a specific time
await ipfs.files.touch('/path/to/file.txt', {
  mtime: new Date('May 23, 2014 14:45:14 -0700')
})

Remove a file or directory.

Where:

• paths are one or more paths to remove
• options is an optional Object that might contain the following keys:
  • recursive is a Boolean value to decide whether or not to remove directories recursively (default: false)

Returns

Example:

// To remove a file
await ipfs.files.rm('/my/beautiful/file.txt')

// To remove multiple files
await ipfs.files.rm('/my/beautiful/file.txt', '/my/other/file.txt')

// To remove a directory
await ipfs.files.rm('/my/beautiful/directory', { recursive: true })

Read a file

Where:

• path is the path of the file to read and must point to a file (and not a directory). It might be:
  • An existing MFS path to a file (e.g. /my-dir/a.txt)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• options is an optional Object that might contain the following keys:
  • offset is an Integer with the byte offset to begin reading from (default: 0)
  • length is an Integer with the maximum number of bytes to read (default: Read to the end of stream)

Returns

Example:

const chunks = []

for await (const chunk of ipfs.files.read('/hello-world')) {
  chunks.push(chunk)
}

console.log(Buffer.concat(chunks).toString())
// Hello, World!

Write to a file.

Where:

• path is the path of the file to write
• content can be:
  • a Buffer
  • an AsyncIterable (note: Node.js readable streams are iterable)
  • a Blob (caveat: will only work in the browser)
  • a string path to a file (caveat: will only work in Node.js)
• options is an optional Object that might contain the following keys:
  • offset is an Integer with the byte offset to begin writing at (default: 0)
  • create is a Boolean to indicate to create the file if it doesn't exist (default: false)
  • truncate is a Boolean to indicate if the file should be truncated after writing all the bytes from content (default: false)
  • parents is a Boolean value to decide whether or not to make the parent directories if they don't exist (default: false)
  • length is an Integer with the maximum number of bytes to read (default: Read all bytes from content)
  • rawLeaves: if true, DAG leaves will contain raw file data and not be wrapped in a protobuf (boolean, default false)
  • cidVersion: the CID version to use when storing the data (storage keys are based on the CID, including its version) (integer, default 0)
  • mode: optional UnixFS mode to create or update the file with - a number or a string that will be interpreted as a base 8 number
  • mtime: A Date object, an object with { sec, nsecs } properties or the output of process.hrtime() or process.hrtime.bigint()

Returns

Example:

await ipfs.files.write('/hello-world', Buffer.from('Hello, world!'))

Move files.

Where:

• from is the path(s) of the source to move
• to is the path of the destination to move to
• options is an optional Object that might contain the following keys:
  • parents is a Boolean value to decide whether or not to make the parent directories if they don't exist (default: false)
  • hashAlg is which algorithm to use when creating CIDs for newly created directories (default: sha2-256) The list of all possible values
  • flush is a Boolean value to decide whether or not to immediately flush MFS changes to disk (default: true)

If from has multiple values then to must be a directory.

If from has a single value and to exists and is a directory, from will be moved into to.

If from has a single value and to exists and is a file, from must be a file and the contents of to will be replaced with the contents of from otherwise an error will be returned.

If from is an IPFS path, and an MFS path exists with the same name, the IPFS path will be chosen.

If from is an IPFS path and the content does not exist in your node's repo, only the root node of the source file with be retrieved from the network and linked to from the destination. The remainder of the file will be retrieved on demand.

All values of from will be removed after the operation is complete unless they are an IPFS path.

Returns

Example:

await ipfs.files.mv('/src-file', '/dst-file')

await ipfs.files.mv('/src-dir', '/dst-dir')

await ipfs.files.mv('/src-file1', '/src-file2', '/dst-dir')

Flush a given path's data to the disk

Where:

• path is an optional string path to flush (default: /)

Returns

Example:

const cid = await ipfs.files.flush('/')

List directories in the local mutable namespace.

Where:

• path is an optional string to show listing for (default: /). It might be:
  • An existing MFS path to a directory (e.g. /my-dir)
  • An IPFS path (e.g. /ipfs/QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks)
  • A CID instance (e.g. new CID('QmWGeRAEgtsHW3ec7U4qW2CyVy7eA2mFRVbk1nb24jFyks'))
• options is an optional Object that might contain the following keys:
  • sort is a Boolean value. If true entries will be sorted by filename (default: false)

Returns

Each object contains the following keys:

• name which is the file's name
• type which is the object's type (directory or file)
• size the size of the file in bytes
• cid the hash of the file (A CID instance)
• mode the UnixFS mode as a Number
• mtime an objects with numeric secs and nsecs properties

Example:

for await (const file of ipfs.files.ls('/screenshots')) {
  console.log(file.name)
}
// 2018-01-22T18:08:46.775Z.png
// 2018-01-22T18:08:49.184Z.png