typestat

#!/bin/sh

listcorpus () {
    find -L corpus \
	 -type f \
	 -regextype posix-extended -regex '.*/(pre|post)(inst|rm)'
}

echo -n "Number of (pre|post)|(rm|inst): "
listcorpus | wc -l

echo "Types found (according to find): "
echo "----------------------------------------------------------------------"
listcorpus | xargs file -b | sort | uniq
echo "----------------------------------------------------------------------"

listcorpus | xargs file -b > .filetypes
echo -n "Number of Posix shell scripts: "
echo -n "(according to file:) "
cat .filetypes |  grep -c "POSIX shell script"

echo -n "Number of Perl scripts "
echo -n "(according to file:) "
cat .filetypes | grep -c "Perl script"

echo -n "Number of Bash scripts "
echo -n "(according to file:) "
cat .filetypes | grep -c "Bourne-Again shell script"

echo -n "Number of Dash scripts "
echo -n "(according to file:) "
cat .filetypes | grep -c "/bin/dash"

echo -n "Number of ELF executables "
echo -n "(according to file:) "
cat .filetypes | grep -c "ELF"

echo -n "Number of ASCII text "
echo -n "(according to file:) "
cat .filetypes |  grep -c -x "ASCII text"

rm .filetypes