How to find files by file type?

"File types" on a Unix system are things like regular files, directories, named pipes, character special files, symbolic links etc. These are the type of files that find can filter on with its -type option.

The find utility can not by itself distinguish between a "shell script", "JPEG image file" or any other type of regular file. These types of data may however be distinguished by the file utility, which looks at particular signatures within the files themselves to determine their type.

A common way to label the different types of data files is by their MIME type, and file is able to determine the MIME type of a file.

Using file with find to detect the MIME type of regular files, and use that to only find shell scripts:

find . -type f -exec sh -c '
    case $( file -bi "$1" ) in
        */x-shellscript*) exit 0
    esac
    exit 1' sh {} ';' -print

or, using bash,

find . -type f \
    -exec bash -c '[[ "$( file -bi "$1" )" == */x-shellscript* ]]' bash {} ';' \
    -print

Add -name sunrise before the -exec if you wish to only detect scripts with that name.

The find command above will find all regular files in or below the current directory, and for each such file call a short in-line shell script. This script runs file -bi on the found file and exits with a zero exit status if the output of that command contains the string /x-shellscript. If the output does not contain that string, it exits with a non-zero exit status which causes find to continue immediately with the next file. If the file was found to be a shell script, the find command will proceed to output the file's pathname (the -print at the end, which could also be replaced by some other action).

The file -bi command will output the MIME type of the file. For a shell script on Linux (and most other systems), this would be something like

text/x-shellscript; charset=us-ascii

while on systems with a slightly older variant of the file utility, it may be

application/x-shellscript

The common bit is the /x-shellscript substring.

Note that on macOS, you would have to use file -bI instead of file -bi because of reasons (the -i option does something quite different). The output on macOS is similar to that of a Linux system.

Would you want to perform some custom action on each found shell script, you could do that with another -exec in place of the -print in the find commands above, but it would also be possible to do

find . -type f -exec sh -c '
    for pathname do
        case $( file -bi "$pathname" ) in
            */x-shellscript*) ;;
            *) continue
        esac

        # some code here that acts on "$pathname"

    done' sh {} +

or, with bash,

find . -type f -exec bash -c '
    for pathname do
        [[ "$( file -bi "$pathname" )" != */x-shellscript* ]] && continue

        # some code here that acts on "$pathname"

    done' bash {} +

Related:

  • Understanding the -exec option of `find`

Tags:

Find

Files

File Command

File Types

Recent Posts