#!/bin/bash

# Extract all parameter entity names declared by a DTD
#
# Not guaranteed to do it all right:
#    Doesn't handle parameter entities, marked sections, comments.
#    Doesn't handle any external files included
#       (to get these, list them on the command line)
#
# Steven J. DeRose 2006-04-18, sderose\@acm.org, sderose@acm.org

# tr -s '\011 '                          # normalize whitespace

if [ "$1" = "-h" ]; then
	echo "Extract all parameter entity names declared by a DTD."
	exit
fi

cat $* |\
sed 's/(<.ENTITY[^>]+)\n/\1 /' |\
sed 's/[ \t]*<.ENTITY/<.ENTITY/g' |\
sed "s/\(<.ENTITY\)[ 	]*/\1 /" |\
tr -s '\011' |\
sed 's/[ 	 	]+(/ /g' |\
grep '<.ENTITY %' |\
grep -iv 'math'