How many file types are in a directory

[agentsmith23]

I am not going to try to deceive any one this question is for a homework assignment, so if you're unwilling to help that is ok.

The issue is that I am trying to find out if there is a way to use the find utility to show how many different file types there are in a specific directory. The book I have actually gives the answer to even numbered questions and this one is answered but I want to know how to get to ha answer in the easiest way. The directory that is in question is /usr/bin and the answer the book gives is "Approximately 20". This directory has hundreds of files (I haven't looked at the exact number). Is there a simple way to find the answer or is it a matter of just using the find utility and manually scrolling through and counting the different file types?

Thanks in advance for any possible help.

Here is a link to the questions and I am working on number 10:
http://www.sobell.com/RH6/answers/05.util.ans.even.pdf

[Dutchy]

I think you have misread the question.
They want you to do it with the "file" command and not the "find" command
I quickly hacked the following together which should do what they want:

PHP Code:

TYPES="$(uniq -c <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep -v symbolic; uniq -cw 10 <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep symbolic | cut -d'`' -f1)"; paste <<< "$TYPES ..."; echo -en "\ndifferent types of files: "; echo "$TYPES" | wc -l 


Output:

PHP Code:

[Dutchy@Dutchy-PC ~]$ TYPES="$(uniq -c <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep -v symbolic; uniq -cw 10 <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep symbolic | cut -d'`' -f1)"; paste <<< "$TYPES ..."; echo -en "\ndifferent types of files: "; echo "$TYPES" | wc -l
     30  Bourne-Again shell script
      1  C++ source
      6  ELF 32-bit LSB executable
   1477  ELF 64-bit LSB executable
     97  ELF 64-bit LSB shared object
      1  executable
      1  Palm OS dynamic library data "#!/bin/sh"
     86  Perl script
    263  POSIX shell script
    111  Python script
      2  setgid ELF 64-bit LSB executable
      1  setgid ELF 64-bit LSB shared object
      5  setgid executable
      3  setuid ELF 64-bit LSB executable
      9  setuid ELF 64-bit LSB shared object
      4  setuid executable
      1  setuid setgid ELF 64-bit LSB executable
    260  symbolic link to  ...

different types of files: 18 


[agentsmith23]

Wow that's awesome! And yes I did mean file not find...ooops

Thanks a lot, now I just need to spend some time dissecting what all of that does so i can fully understand it.

[Dutchy]

Glad I could help you out a bit.
The code is quite a quick and dirty solution and might indeed not be easy to understand.
It actually consist out of 3 parts.

First it looks up the file types through the use of magic paterns of all the files in /usr/bin. Then it it cuts out everything behind the "," (comma) and then cuts out everything before the ":" (colon). The mess is nicely sorted so that uniq can make something of it. The output of this all is put into uniq which will look for duplicate lines and merge them so every line is unique, it will also count the duplicates. After that grep throws away all the lines with "symbolic" in it (because that are links to other files and do not play nice with "uniq").

PHP Code:

uniq -c <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep -v symbolic 


Now it will do the same thing again but this time the symbolic links are taken into account (the file output is unique for every symbolic link because it involves the location of the file it points to).

PHP Code:

uniq -cw 10 <<< "$(file -N /usr/bin/* | cut -d, -f1 | cut -d: -f2 | sort)" | grep symbolic | cut -d'`' -f1 


The past two commands are put into the $TYPES variable.
With the paste command the output of $TYPES gets some trailing dots (otherwise it would look a bit weird) and is printed to the screen.

PHP Code:

paste <<< "$TYPES ..." 


Then this will print a blank line (the -e enables the interpretation of special characters (in this case \n, newline) and the -n switch turns the trailing new line off (so the next command can be printed on the same line).

PHP Code:

echo -en "\ndifferent types of files: " 


Lastly the wc (word count) command counts the amount of lines in the $TYPES variable

PHP Code:

echo "$TYPES" | wc -l 


If you bump into trouble again and need some more info on bash you can take a look over here:
http://www.tldp.org/LDP/abs/html/index.html

[glennzo]

How would one expand on this to include subfolders of the specified path?

[george_toolan]

Try the find command ;-)

TYPES="$(uniq -c <<< "$(find /usr/bin | file -N -f - | cut -d, -f1 | cut -d: -f2 | sort)" | grep -v symbolic; uniq -cw 10 <<< "$(find /usr/bin | file -N -f - | cut -d, -f1 | cut -d: -f2 | sort)" | grep symbolic | cut -d'`' -f1)"; paste <<< "$TYPES ..."; echo -en "\ndifferent types of files: "; echo "$TYPES" | wc -l

This will add another file type "directory".

[stevea]

The OPs honesty is refreshing, pleasing and says something of character - welcome agentsmith23
Dutchy make masterful use of bash redirection - kudos - but let me play devil's advocate ....

The result falls short in several respect IMO. If the issue is 'file types' and not 'file metadata categories then we have a different problem than the one answered. 'symlinks' describe metadata, as do 'sticky', 'setuid' ,'setgid' . Also ", with very long lines" is pure baloneum and has no place in determining file type.

So something like ....

# file -bNL -e elf  /usr/bin/* | sed -e 's/setuid //;s/setgid //;s/sticky //;s/, with very long lines//' | sort | uniq -c
     33 Bourne-Again shell script, ASCII text executable
      5 Bourne-Again shell script, UTF-8 Unicode text executable
      1 ELF 32-bit LSB executable, Intel 80386, version 1 (GNU/Linux)
      4 ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV)
     30 ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux)
   1574 ELF 64-bit LSB executable, x86-64, version 1 (SYSV)
      1 ELF 64-bit LSB shared object, x86-64, version 1 (GNU/Linux)
     94 ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)
      1 Palm OS dynamic library data "#!/bin/sh"
     78 Perl script, ASCII text executable
      1 Perl script, ISO-8859 text executable
      1 Perl script, UTF-8 Unicode text executable
    257 POSIX shell script, ASCII text executable
      1 POSIX shell script, ISO-8859 text executable
      9 POSIX shell script, UTF-8 Unicode text executable
     88 Python script, ASCII text executable
      7 Python script, UTF-8 Unicode text executable

'flle' certainly may misidentify 'Palm OS' libs 'C++ source' and likely makes poor distinctions between ASCII, ISO-8859, UTF-8 Unicode, but that's out of scope.

The OS ABI "version 1 (GNU/Linux)" vs "version 1 (SYSV)" is an important distinction of file type and should not be ignored.

[Dutchy]

That one looks indeed way more sane, is better thought through and last but not least it is a lot shorter too

[jpollard]

Guys - save a process... they are rare. If you are going to sort, why not use "sort -u" to eliminate duplicates?

granted, the uniq utility allows counting the number of duplicates.

[Gareth Jones]

Depending how interested you are in the specifics of the types and context, you could simplify it further to something like:

$ file -b --mime /usr/bin/* | sort -u
application/x-executable; charset=binary
application/x-sharedlib; charset=binary
executable, regular file, no read permission
inode/symlink
inode/symlink; charset=binary
regular file, no read permission
text/plain; charset=utf-8
text/x-c++; charset=us-ascii
text/x-perl; charset=iso-8859-1
text/x-perl; charset=us-ascii
text/x-perl; charset=utf-8
text/x-python; charset=us-ascii
text/x-python; charset=utf-8
text/x-ruby; charset=us-ascii
text/x-shellscript; charset=iso-8859-1
text/x-shellscript; charset=us-ascii
text/x-shellscript; charset=utf-8
text/x-tex; charset=us-ascii