Identifying a file with “file” in a shell script?

Question

I'm currently writing a shell script that needs to be able to identify the actual type of files (not their extensions) with the unix "file" utility. However, that tool outputs human readable text which might be somewhat difficult to parse. Before I write something that just matches specific substrings found in its output, is there a better way to approach this problem?

Answer 1

Using the --mime-type argument to file gives MIME type which is easier to parse.

$ file --mime-type /etc/passwd
/etc/passwd: text/plain

The -i option gives a little bit more information (encoding type) if needed:

$ file -i /etc/passwd
/etc/passwd: text/plain; charset=us-ascii