Detect the file type of a file, stream, or data

The file type is detected by checking the magic number of the buffer.

This package is for detecting binary-based file formats, not text-based formats like .txt, .csv, .svg, etc.

We accept contributions for commonly used modern file formats, not historical or obscure ones. Open an issue first for discussion.

Install

npm install file-type

This package is an ESM package. Your project needs to be ESM too. Read more. For TypeScript + CommonJS, see load-esm. If you use it with Webpack, you need the latest Webpack version and ensure you configure it correctly for ESM.

[!IMPORTANT] File type detection is based on binary signatures (magic numbers) and is a best-effort hint. It does not guarantee the file is actually of that type or that the file is valid/not malformed.

Robustness against malformed input is best-effort. When processing untrusted files on a server, enforce a reasonable file size limit and use a worker thread with a timeout (e.g., make-asynchronous). These are not considered security issues in this package.

Usage

Node.js

Determine file type from a file:

import {fileTypeFromFile} from 'file-type';

console.log(await fileTypeFromFile('Unicorn.png'));
//=> {ext: 'png', mime: 'image/png'}

Determine file type from a Uint8Array/ArrayBuffer, which may be a portion of the beginning of a file:

import {fileTypeFromBuffer} from 'file-type';
import {readChunk} from 'read-chunk';

const buffer = await readChunk('Unicorn.png', {length: 4100});

console.log(await fileTypeFromBuffer(buffer));
//=> {ext: 'png', mime: 'image/png'}

Determine file type from a stream:

import {fileTypeFromStream} from 'file-type';

const url = 'https://upload.wikimedia.org/wikipedia/en/a/a9/Example.jpg';

const response = await fetch(url);
const fileType = await fileTypeFromStream(response.body);

console.log(fileType);
//=> {ext: 'jpg', mime: 'image/jpeg'}

API

fileTypeFromBuffer(buffer, options)

Detect the file type of a Uint8Array or ArrayBuffer.

The file type is detected by checking the magic number of the buffer.

If file access is available, it is recommended to use fileTypeFromFile() instead.

Returns a Promise for an object with the detected file type:

• ext - One of the supported file types
• mime - The MIME type

Or undefined when there is no match.

buffer

Type: Uint8Array | ArrayBuffer

A buffer representing file data. It works best if the buffer contains the entire file. It may work with a smaller portion as well.

fileTypeFromFile(filePath, options)

Detect the file type of a file path.

Only available in environments where node:fs is available, such as Node.js. To read from a File, see fileTypeFromBlob().

The file type is detected by checking the magic number of the file.

Returns a Promise for an object with the detected file type:

• ext - One of the supported file types
• mime - The MIME type

Or undefined when there is no match.

filePath

Type: string

The file path to parse.

fileTypeFromStream(stream, options)

Detect the file type of a web ReadableStream.

The file type is detected by checking the magic number of the buffer.

Returns a Promise for an object with the detected file type:

• ext - One of the supported file types
• mime - The MIME type

Or undefined when there is no match.

stream

Type: Web ReadableStream

A readable stream representing file data.

[!TIP] If you have a Node.js stream.Readable, convert it with Readable.toWeb().

fileTypeFromBlob(blob, options)

Detect the file type of a Blob.

[!TIP] A File object is a Blob and can be passed in here.

It will stream the underlying Blob.

The file type is detected by checking the magic number of the blob.

Returns a Promise for an object with the detected file type:

• ext - One of the supported file types
• mime - The MIME type

Or undefined when there is no match.

import {fileTypeFromBlob} from 'file-type';

const blob = new Blob(['<?xml version="1.0" encoding="ISO-8859-1" ?>'], {
	type: 'text/plain',
	endings: 'native'
});

console.log(await fileTypeFromBlob(blob));
//=> {ext: 'xml', mime: 'application/xml'}

blob

Type: Blob

fileTypeFromTokenizer(tokenizer, options)

Detect the file type from an ITokenizer source.

This method is used internally, but can also be used for a special "tokenizer" reader.

A tokenizer propagates the internal read functions, allowing alternative transport mechanisms, to access files, to be implemented and used.

Returns a Promise for an object with the detected file type:

• ext - One of the supported file types
• mime - The MIME type

Or undefined when there is no match.

An example is @tokenizer/http, which requests data using HTTP-range-requests. A difference with a conventional stream and the tokenizer, is that it can ignore (seek, fast-forward) in the stream. For example, you may only need and read the first 6 bytes, and the last 128 bytes, which may be an advantage in case reading the entire file would take longer.

import {makeTokenizer} from '@tokenizer/http';
import {fileTypeFromTokenizer} from 'file-type';

const audioTrackUrl = 'https://test-audio.netlify.com/Various%20Artists%20-%202009%20-%20netBloc%20Vol%2024_%20tiuqottigeloot%20%5BMP3-V2%5D/01%20-%20Diablo%20Swing%20Orchestra%20-%20Heroines.mp3';

const httpTokenizer = await makeTokenizer(audioTrackUrl);
const fileType = await fileTypeFromTokenizer(httpTokenizer);

console.log(fileType);
//=> {ext: 'mp3', mime: 'audio/mpeg'}

Or use @tokenizer/s3 to determine the file type of a file stored on Amazon S3:

import {S3Client} from '@aws-sdk/client-s3';
import {makeChunkedTokenizerFromS3} from '@tokenizer/s3';
import {fileTypeFromTokenizer} from 'file-type';

// Initialize S3 client
const s3 = new S3Client();

// Initialize the S3 tokenizer.
const s3Tokenizer = await makeChunkedTokenizerFromS3(s3, {
	Bucket: 'affectlab',
	Key: '1min_35sec.mp4'
});

// Figure out what kind of file it is.
const fileType = await fileTypeFromTokenizer(s3Tokenizer);
console.log(fileType);

Note that only the minimum amount of data required to determine the file type is read (okay, just a bit extra to prevent too many fragmented reads).

tokenizer

Type: ITokenizer

A file source implementing the tokenizer interface.

fileTypeStream(webStream, options?)

Returns a Promise which resolves to the original readable stream argument, but with an added fileType property, which is an object like the one returned from fileTypeFromFile().

This method can be handy to put in a stream pipeline, but it comes with a price. Internally stream() builds up a buffer of sampleSize bytes, used as a sample, to determine the file type. The sample size impacts the file detection resolution. A smaller sample size will result in lower probability of the best file type detection.

webStream

Type: Web ReadableStream

The input stream.

options

Type: object

Supports the following options in addition to the general options:

sampleSize

Type: number
Default: 4100

The sample size in bytes.

Example

import {fileTypeStream} from 'file-type';

const url = 'https://upload.wikimedia.org/wikipedia/en/a/a9/Example.jpg';

const response = await fetch(url);
const stream = await fileTypeStream(response.body, {sampleSize: 1024});

if (stream.fileType?.mime === 'image/jpeg') {
	// stream can be used to stream the JPEG image (from the very beginning of the stream)
}

supportedExtensions

Returns a Set<string> of file extensions that can be detected.

supportedMimeTypes

Returns a Set<string> of MIME types that can be detected.

Options

customDetectors

Array of custom file type detectors to run before default detectors.

For example:

import {fileTypeFromFile} from 'file-type';
import {detectXml} from '@file-type/xml';

const fileType = await fileTypeFromFile('sample.kml', {customDetectors: [detectXml]});
console.log(fileType);

maxBufferSize

Type: number
Default: Infinity

Maximum size in bytes of an individual buffer allocated to read file content, or a stream lookahead request. Must be a positive safe integer or Infinity.

File content can determine how much data a detector requests, for example through a field specifying the length of a metadata entry. This opt-in limit bounds the memory allocations needed to fulfill those reads, including reads made by custom detectors. Setting a finite cap can help prevent excessive read-buffer allocations driven by untrusted file content (CWE-789: Memory Allocation with Excessive Size Value). Built-in reads already have format-specific bounds; choose a finite cap to suit your application. An oversized request rejects detection with strtok3.BufferSizeError, which exposes requestedSize and maxBufferSize. During gzip content probing, this error is suppressed and detection falls back to gz; the cap still prevents the oversized allocation.

const fileType = await fileTypeFromFile('sample.bin', {maxBufferSize: 64 * 1024 * 1024});

This is not a total memory limit. It does not cover buffers allocated directly by detectors or fileTypeStream() sampling. When using fileTypeFromTokenizer(), configure maxBufferSize on the input reader itself; this option does not override that reader’s limits. The built-in format-specific limits, including ZIP text-entry limits, apply independently of this option.

For FileTypeParser.toDetectionStream(), configure detection options such as maxBufferSize through the parser constructor. The method only accepts stream options such as sampleSize.

mpegOffsetTolerance

Default: 0

Specifies the byte tolerance for locating the first MPEG audio frame (e.g. .mp1, .mp2, .mp3, .aac).

Allows detection to handle slight sync offsets between the expected and actual frame start. Common in malformed or incorrectly muxed files, which, while technically invalid, do occur in the wild.

A tolerance of 10 bytes covers most cases.

Custom detectors

Custom file type detectors are plugins designed to extend the default detection capabilities. They allow support for uncommon file types, non-binary formats, or customized detection behavior.

Detectors can be added via the constructor options or by directly modifying FileTypeParser#detectors. Detectors provided through the constructor are executed before the default ones.

[!NOTE] Not safe for concurrent use. Create a new instance per call if you need concurrency.

Example adding a detector

import {FileTypeParser} from 'file-type';
import {detectXml} from '@file-type/xml';

const parser = new FileTypeParser({customDetectors: [detectXml]});
const fileType = await parser.fromFile('sample.kml');
console.log(fileType);

Available third-party file-type detectors

• @file-type/av: Improves detection of audio and video file formats, with accurate differentiation between the two
• @file-type/cfbf: Detects Compound File Binary Format (CFBF) based formats, such as Office 97–2003 documents and .msi.
• @file-type/pdf: Detects PDF based file types, such as Adobe Illustrator
• @file-type/xml: Detects common XML file types, such as GLM, KML, MusicXML, RSS, SVG, and XHTML

Detector execution flow

If a detector returns undefined, the following rules apply:

1. No Tokenizer Interaction: If the detector does not modify the tokenizer's position, the next detector in the sequence is executed.
2. Tokenizer Interaction: If the detector modifies the tokenizer's position (tokenizer.position is advanced), no further detectors are executed. In this case, the file type remains undefined, as subsequent detectors cannot evaluate the content. This is an exceptional scenario, as it prevents any other detectors from determining the file type.

Writing your own custom detector

Below is an example of a custom detector. This can be passed to the FileTypeParser via the customDetectors option.

import {FileTypeParser} from 'file-type';

const unicornDetector = {
	id: 'unicorn', // May be used to recognize the detector in the detector list
	async detect(tokenizer) {
		const unicornHeader = [85, 78, 73, 67, 79, 82, 78]; // "UNICORN" in ASCII decimal

		const buffer = new Uint8Array(unicornHeader.length);
		await tokenizer.peekBuffer(buffer, {length: unicornHeader.length, mayBeLess: true});
		if (unicornHeader.every((value, index) => value === buffer[index])) {
			return {ext: 'unicorn', mime: 'application/unicorn'};
		}

		return undefined;
	}
}

const buffer = new Uint8Array([85, 78, 73, 67, 79, 82, 78]);
const parser = new FileTypeParser({customDetectors: [unicornDetector]});
const fileType = await parser.fromBuffer(buffer);
console.log(fileType); // {ext: 'unicorn', mime: 'application/unicorn'}

/**
@param tokenizer - The [tokenizer](https://github.com/Borewit/strtok3#tokenizer) used to read file content.
@param fileType - The file type detected by standard or previous custom detectors, or `undefined` if no match is found.
@returns The detected file type, or `undefined` if no match is found.
*/
export type Detector = {
	id: string;
	detect: (tokenizer: ITokenizer, fileType?: FileTypeResult) => Promise<FileTypeResult | undefined>;
};

Abort signal

Some async operations can be aborted by passing an AbortSignal to the FileTypeParser constructor.

import {FileTypeParser} from 'file-type';

const abortController = new AbortController();

const parser = new FileTypeParser({signal: abortController.signal});

const promise = parser.fromStream(blob.stream());

abortController.abort(); // Abort file-type reading from the Blob stream.

Supported file types

MIME media subtypes prefixed with x-ft- are custom and defined by us. They are neither formally registered with IANA nor based on any informal conventions.

• 3g2 - Multimedia container format defined by the 3GPP2 for 3G CDMA2000 multimedia services
• 3gp - Multimedia container format defined by the Third Generation Partnership Project (3GPP) for 3G UMTS multimedia services
• 3mf - 3D Manufacturing Format
• 7z - 7-Zip archive
• Z - Unix Compressed File
• aac - Advanced Audio Coding
• ac3 - ATSC A/52 Audio File
• ace - ACE archive
• aif - Audio Interchange file
• alias - macOS Alias file
• amr - Adaptive Multi-Rate audio codec
• ape - Monkey's Audio
• apk - Android package format
• apng - Animated Portable Network Graphics
• ar - Archive file
• arj - Archive file
• arrow - Columnar format for tables of data
• arw - Sony Alpha Raw image file
• asar - Archive format primarily used to enclose Electron applications
• asf - Advanced Systems Format
• avi - Audio Video Interleave file
• avif - AV1 Image File Format
• avro - Object container file developed by Apache Avro
• blend - Blender project
• bmp - Bitmap image file
• bpg - Better Portable Graphics file
• bz2 - Archive file
• cab - Cabinet file
• caf - Apple Core Audio Format
• cfb - Compound File Binary Format
• chm - Microsoft Compiled HTML Help
• class - Java class file
• cpio - Cpio archive
• cr2 - Canon Raw image file (v2)
• cr3 - Canon Raw image file (v3)
• crx - Google Chrome extension
• cur - Icon file
• dat - Windows registry hive file
• dcm - DICOM Image File
• deb - Debian package
• dff - Direct Stream Digital Interchange File Format (DSDIFF)
• dmg - Apple Disk Image
• dng - Adobe Digital Negative image file
• docm - Microsoft Word macro-enabled document
• docx - Microsoft Word document
• dotm - Microsoft Word macro-enabled template
• dotx - Microsoft Word template
• drc - Google's Draco 3D Data Compression
• dsf - Sony DSD Stream File (DSF)
• dwg - Autodesk CAD file
• elf - Unix Executable and Linkable Format
• eot - Embedded OpenType font
• eps - Encapsulated PostScript
• epub - E-book file
• exe - Executable file
• f4a - Audio-only ISO base media file format used by Adobe Flash Player
• f4b - Audiobook and podcast ISO base media file format used by Adobe Flash Player
• f4p - ISO base media file format protected by Adobe Access DRM used by Adobe Flash Player
• f4v - ISO base media file format used by Adobe Flash Player
• fbx - Filmbox is a proprietary file format used to provide interoperability between digital content creation apps.
• flac - Free Lossless Audio Codec
• flif - Free Lossless Image Format
• flv - Flash video
• gif - Graphics Interchange Format
• glb - GL Transmission Format
• gz - Archive file
• heic - High Efficiency Image File Format
• icc - ICC Profile
• icns - Apple Icon image
• ico - Windows icon file
• ics - iCalendar
• indd - Adobe InDesign document
• iso - ISO 9660 disc image
• it - Audio module format: Impulse Tracker
• j2c - JPEG 2000
• jar - Java archive
• jls - Lossless/near-lossless compression standard for continuous-tone images
• jmp - JMP data file format by SAS Institute
• jp2 - JPEG 2000
• jpg - Joint Photographic Experts Group image
• jpm - JPEG 2000
• jpx - JPEG 2000
• jxl - JPEG XL image format
• jxr - Joint Photographic Experts Group extended range
• key - Apple Keynote presentation
• ktx - OpenGL and OpenGL ES textures
• lnk - Microsoft Windows file shortcut
• lz - Archive file
• lz4 - Compressed archive created by one of a variety of LZ4 compression utilities
• lzh - LZH archive
• m4a - Audio-only MPEG-4 files
• m4b - Audiobook and podcast MPEG-4 files, which also contain metadata including chapter markers, images, and hyperlinks
• m4p - MPEG-4 files with audio streams encrypted by FairPlay Digital Rights Management as were sold through the iTunes Store
• m4v - Video container format developed by Apple, which is very similar to the MP4 format
• macho - Mach-O binary format
• mid - Musical Instrument Digital Interface file
• mie - Dedicated meta information format which supports storage of binary as well as textual meta information
• mj2 - Motion JPEG 2000
• mkv - Matroska video file
• mobi - Mobipocket
• mov - QuickTime video file
• mp1 - MPEG-1 Audio Layer I
• mp2 - MPEG-1 Audio Layer II
• mp3 - Audio file
• mp4 - MPEG-4 Part 14 video file
• mpc - Musepack (SV7 & SV8)
• mpg - MPEG-1 file
• mts - MPEG-2 Transport Stream, both raw and Blu-ray Disc Audio-Video (BDAV) versions
• mxf - Material Exchange Format
• nef - Nikon Electronic Format image file
• nes - Nintendo NES ROM
• numbers - Apple Numbers spreadsheet
• odg - OpenDocument for drawing
• odp - OpenDocument for presentations
• ods - OpenDocument for spreadsheets
• odt - OpenDocument for word processing
• oga - Audio file
• ogg - Audio file
• ogm - Audio file
• ogv - Audio file
• ogx - Audio file
• opus - Audio file
• orf - Olympus Raw image file
• otf - OpenType font
• otg - OpenDocument template for drawing
• otp - OpenDocument template for presentations
• ots - OpenDocument template for spreadsheets
• ott - OpenDocument template for word processing
• pages - Apple Pages document
• parquet - Apache Parquet
• pcap - Libpcap File Format
• pdf - Portable Document Format
• pgp - Pretty Good Privacy
• png - Portable Network Graphics
• potm - Microsoft PowerPoint macro-enabled template
• potx - Microsoft PowerPoint template
• ppsm - Office PowerPoint 2007 macro-enabled slide show
• ppsx - Office PowerPoint 2007 slide show
• pptm - Microsoft PowerPoint macro-enabled document
• pptx - Microsoft PowerPoint document
• ps - PostScript
• psd - Adobe Photoshop document
• pst - Personal Storage Table file
• qcp - Tagged and chunked data
• raf - Fujifilm RAW image file
• rar - Archive file
• reg - Windows registry (entries) file format
• rm - RealMedia
• rpm - Red Hat Package Manager file
• rtf - Rich Text Format
• rw2 - Panasonic RAW image file
• s3m - Audio module format: ScreamTracker 3
• sav - SPSS Statistical Data File
• shp - Geospatial vector data format
• skp - SketchUp
• spx - Audio file
• sqlite - SQLite file
• stl - Standard Tessellated Geometry File Format (ASCII only)
• swf - Adobe Flash Player file
• tar - Tape archive or tarball
• tar.gz - Gzipped tape archive (tarball)
• tif - Tagged Image file
• ttc - TrueType Collection font
• ttf - TrueType font
• vcf - vCard
• voc - Creative Voice File
• vsdx - Microsoft Visio File
• vtt - WebVTT File (for video captions)
• wasm - WebAssembly intermediate compiled format
• wav - Waveform Audio file
• webm - Web video file
• webp - Web Picture format
• woff - Web Open Font Format
• woff2 - Web Open Font Format
• wv - WavPack
• xcf - eXperimental Computing Facility
• xlsm - Microsoft Excel macro-enabled document
• xlsx - Microsoft Excel document
• xltm - Microsoft Excel macro-enabled template
• xltx - Microsoft Excel template
• xm - Audio module format: FastTracker 2
• xml - eXtensible Markup Language
• xpi - XPInstall file
• xz - Compressed file
• zip - Archive file
• zst - Archive file

ISO 9660 requires random-access input or fileTypeStream() with sampleSize >= 32_774; fileTypeFromStream() is unsupported.

When [Content_Types].xml cannot be reached — it may trail the content entries, and only so much of what precedes it is scanned — the type is guessed from the archive's directory names: word/, xl/ and ppt/ are reported as docx, xlsx and pptx. Macro-enabled and template variants are therefore reported as their base type, and a non-OOXML archive containing one of those directories is reported as OOXML.

Pull requests are welcome for additional commonly used file types.

The following file types will not be accepted, but most of them are supported by third-party detectors.

• MS-CFB: Microsoft Compound File Binary File Format based formats
  • .doc - Microsoft Word 97-2003 Document
  • .xls - Microsoft Excel 97-2003 Document
  • .ppt - Microsoft PowerPoint 97-2003 Document
  • .msi - Microsoft Windows Installer
• .csv - Reason.
• .svg

Related

• file-type-cli - CLI for this module
• image-dimensions - Get the dimensions of an image

Maintainers

• Sindre Sorhus
• Borewit