detect-file-type

Detect file type by signatures. Fork of detect-file-type.

• Removed dependency on node's fs module to enable this package to be used on browsers
• Increase performance by removing usage of text recoding
• Typescript type definition file included

Supported types

jpg, png, gif, webp, flif, cr2, tif, bmp, jxr, psd, zip, epub, xpi, tar, rar, gz, bz2, 7z, dmg, mov, mp4, m4v, m4a, 3g2, 3gp, avi, wav, qcp, mid, mkv, webm, wasm, asf, wmv, wma, mpg, mp3, opus, ogg, ogv, oga, ogm, ogx, spx, flac, ape, wv, amr, pdf, exe, swf, rtf, woff, woff2, eot, ttf, otf, ico, cur, flv, ps, xz, sqlite, nes, dex, crx, elf, cab, deb, ar, rpm, Z, lz, msi, mxf, mts, blend, bpg, jp2, jpx, jpm, mj2, aif, xml, svg, mobi, heic, ktx, dcm, mpc, ics, glb, pcap, html, dll, xlsx, xlsm, xlsb, xlam, docx, docm, dotm, dotx, pptx, pptm, ppsm, ppsx

Installation

  npm i --save detect-file-type-lite

Usage

1  import { readFile } from 'fs/promises'
2  import FileTypeDetector from 'detect-file-type-lite'
3
4  const fileTypeDetector = new FileTypeDetector()
5  const buffer = await readFile('img.jpg')
6  const result = await fileTypeDetector.fromBuffer(buffer);
7  console.log(result.ext)  // jpg

API

fromBuffer(buffer) => Promise<FileTypeResult | null>

Detect file type from buffer

• buffer - uint8array/Buffer

addSignature(signature) => void

Add new signature for file type detecting

• signature - a signature. See section about it below

addCustomFunction(fn) => void

Add custom function which receive buffer and trying to detect file type.

• fn - function which receive buffer

This method needed for more complicated cases like DLL for example. A DLL file will be detected as a normal EXE file, unless you perform deeper checks on its headers.

1import { readFile } from 'fs/promises'
2import FileTypeDetector from 'detect-file-type-lite'
3const fileTypeDetector = new FileTypeDetector()
4
5fileTypeDetector.addCustomFunction((buffer: Buffer) => {
6  if(buffer.length < 2) return false
7  
8  const mzHeader = Buffer.from([0x4D, 0x5A])
9  if(buffer.compare(mzHeader, 0, mzHeader.length, 0, mzHeader.length) != 0) return false
10
11  const PE_HEADER_OFFSET = buffer.readUInt32LE(0x3c)
12  const CHARACTERISTICS_OFFSET = PE_HEADER_OFFSET + 0x16
13  const characteristics = buffer.readUInt16LE(CHARACTERISTICS_OFFSET)
14  // IMAGE_FILE_DLL == 0x2000
15  if((characteristics & 0x2000) === 0x2000) {
16    return {
17      ext: 'dll',
18      mime: 'application/octet-stream'
19    }
20  }
21  return false
22})
23
24const buf = await readFile('winrsmgr.dll')
25const res = await fileTypeDetector.fromBuffer(buf)
26console.log(res)  // { ext: 'dll', mime: 'application/octet-stream' }

Some custom functions, including the above example on DLL file type detection, are already declared during object instantiation. See custom-functions.ts for more details.

Custom functions can be async.

Signature and creating your own signatures

Detecting of file type work via signatures. The simplest signature in JSON format looks like:

1{
2  "type": "jpg",
3  "ext": "jpg",
4  "mime": "image/jpeg",
5  "rules": [
6    { "type": "equal", "start": 0, "end": 2, "bytes": "ffd8"  }
7  ]
8}

params:

• type - signature type, mostly it's the same as param 'ext'
• ext - file extension
• iana - optional iana registered mime type - will be added when actual used mime differs from iana, or when the old mime type we used was wrong
• mime - mime type of file
• rules - list of rules for detecting

More details about param rules:

This param have to be array of objects

• type - a rule type. There are available a few types: equal, notEqual, contains, notContains, or, and, default
• search - a searching rule, that keeps the offset of the searched bytes in a special id.
• search_ref - a reference to a previously performed search, start and end will be offset by it.

More details about: equal, notEqual, contains & notContains.

• equal - here is required field bytes. We get a dump of buffer from start (equals 0 by default) to end (equals buffer.length by default). After that we compare the dump with value in param bytes. If values are equal then this rule is correct.
• notEqual - here is required field bytes. We get a dump of buffer from start (equals 0 by default) to end (equals buffer.length by default). After that we compare the dump with value in param bytes. If values aren't equal then this rule is correct.
• contains - here is required field bytes. We get a dump of buffer from start (equals 0 by default) to end (equals buffer.length by default). After that we try to find the sequence from bytes in this dump. If the dump contains bytes then rules is correct.
• notContains - here is required field bytes. We get a dump of buffer from start (equals 0 by default) to end (equals buffer.length by default). After that we try to find the sequence from bytes in this dump. If the dump contains bytes then rules is incorrect.

More details about the rule types or and and

Actually, these types are necessary when you work with complicated signatures. For example, when file contains few sequences of bytes in different parts of file. Here is required field 'rules', where you should define set of another rules. See example:

1{
2  "type": "tif",
3  "ext": "tif",
4  "mime": "image/tiff",
5  "rules": [
6    { "type": "and", "rules":
7      [
8        { "type": "notEqual", "start": 8, "end": 10, "bytes": "4352" },
9        { "type": "or", "rules":
10            [
11             { "type": "equal", "start": 0, "end": 4, "bytes": "49492a00" },
12             { "type": "equal", "start": 0, "end": 4,  "bytes": "4d4d002a" }
13           ]
14          }
15      ]
16    }
17   ]
18 }

Explanation: If dump starts from 8th byte and ends to 10th byte isn't equal "4352", and dump starts from 0 and ends to 4th byte is equal "49492a00" or is equal "4d4d002a" then data looks like file with 'tif' format.

• or - means that any rules from rules should be correct. If at least 1 rule is correct then list are correct too.
• and - means that each rule from rules should be correct. If all rules are correct then list is correct. When at least 1 rule fail then all list is incorrect.

The rules or and and can be nested without restrictions.

The default type is special and is used as a fallback when a set of or rules did not match, inside a larger tree with multiple mime detections.

More details about the search object

• id - id to assign to the result (reference later with search_ref)
• start/end - range to search in
• bytes - bytes to search for

License

WTFPL © Dmitry Pavlovsky