Is there a java library equivalent to file command in unix

Question

Is there any java library that is similar to unix's command file?

ie:

$ file somepicture.png
somepicture.png PNG image, 805 x 292, 8-bit/color RGB, non-interlaced

The file command is such a nice tool. I need something that can tell me if the file is really what I want it to be. (ie a picture, document etc)

I know I can run the command file, but I am looking for a java library, not running the actual unix command.

I ended up using Apache Tika. JMimeMagic was poorly documented. (Not that Tika was better, but the API was easier IMO) — Shervin Asgari, May 18 '10 at 12:45

Hasturkun · Accepted Answer · 2016-05-26T06:53:55.690

18

A quick google search (for the admittedly non-obvious) "java magic file detection" brings up a fairly nice looking article, "Get the Mime Type from a File" which suggests you use one of the following:

edited May 26 '16 at 06:53

answered Apr 28 '10 at 11:46

Hasturkun

35,395
6
71
104

Thanks. I think one of these two should do the trick. Its really documents and images I need. – Shervin Asgari Apr 28 '10 at 17:28

score 11 · Answer 2 · edited Jan 27 '17 at 18:38

11

Since Java 1.7 you can use Files.probeContentType() to probe a file. Out of box it uses the mechanism on the platform to guess the content type, or you can plugin your own detector if you want.

edited Jan 27 '17 at 18:38

Aubin

14,617
9
61
84

answered May 05 '11 at 12:59

Alan

2,807
3
22
6

score 3 · Answer 3 · answered Apr 28 '10 at 11:43

3

You could look at jmimemagic (tutorial). We've been using it for a while to validate uploaded images. No problems so far.

answered Apr 28 '10 at 11:43

sfussenegger

35,575
15
95
119

score 2 · Answer 4 · answered Apr 28 '10 at 11:47

2

I am not sure it is exactly what you are looking for, but the following link can maybe help you :

http://www.rgagnon.com/javadetails/java-0487.html

answered Apr 28 '10 at 11:47

Laurent

14,122
13
57
89

+1 for giving the JDK-only solution. I'm not sure how good that method is, but it could be what the user is looking for. – Brett Kail Apr 28 '10 at 13:37
2

@bkail `javax.activation.MimetypesFileTypeMap` only checks the file extension which surely isn't a reliable way of determining "if the file is really what I want it to be" – sfussenegger Apr 28 '10 at 14:15
Yes the JDK solution is not working like `file` in unix. `file` really looks at the file and tells what it is. Doesn't care about extension like windows does – Shervin Asgari Apr 28 '10 at 17:26
@sfussenegger java.net.URLConnection.guessContentTypeFromStream(InputStream) surely does not use file extensions to make its determination. – Brett Kail Apr 29 '10 at 14:21
@bkail That wasn't mentioned in the linked article, was it? Anyway, it's not working reliably. I've just tried and it failed for a simple jpg image. – sfussenegger Apr 29 '10 at 14:53
@sfussenegger Huh, I thought getContent/getContentType used guessContentTypeFromStream. Ah, apparently only for "jar:" URLs. Thanks for the clarification... – Brett Kail Apr 29 '10 at 15:55

score 1 · Answer 5 · answered Apr 28 '10 at 12:09

1

Have a look at mime-utils. It works with content and/or with extensions.

answered Apr 28 '10 at 12:09

openCage

2,735
1
18
24

score 1 · Answer 6 · answered May 18 '10 at 12:51

1

The closest thing in the JDK is URLConnection.guessContentTypeFromStream

answered May 18 '10 at 12:51

finnw

47,861
24
143
221

score 0 · Answer 7 · answered Jan 21 '19 at 15:10

0

Was looking for the same thing and found this: https://github.com/j256/simplemagic - it seems to be a clone of 'file', it even uses a copy of files built-in magic file.

answered Jan 21 '19 at 15:10

Markus Duft

151
1
5

Is there a java library equivalent to file command in unix

7 Answers7

Linked