trivial-validate

#!/bin/sh
# This is a shell script that looks for common problems and typos in rules.

set -o errexit -o noclobber -o nounset

[ -n "$1" ] && cd "$1"

if [ "$(echo *.xml)" = '*.xml' ]
then
   echo There are no readable XML files in the current or specified directory.
   echo Run this script inside a directory containing XML rule files or else
   echo specify such a directory on the command line.  E.g.,
   echo
   echo "$0" src/chrome/content/rules
   exit 2
fi

echo "-- Rules not anchored to beginning of a line:"
if grep from= *.xml | cut -d\" -f2 | grep '^[^^]'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules with unescaped dots outside of brackets and left of a slash:"
if grep from= *.xml | cut -d\" -f2 | sed 's/\[[^]]*\]//g' | egrep 'http.?.?://[^/]*[^\]\.[^*]'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules with backslashes in to pattern:"
if grep 'to="' *xml | sed 's/^.*to="//' | sed 's/\".*$//' | grep '\\'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules not containing trailing slash in from pattern:"
if grep from= *.xml | cut -d\" -f2 | grep -v '//.*/'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules not containing trailing slash in to pattern:"
if grep 'to="' *xml | sed 's/^.*to="//' | sed 's/\".*$//' | grep -v '//.*/'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules with missing closing slash in rule XML tag:"
if grep to= *xml | grep '[^/]>'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules with multiple wildcards in a single target rule:"
if grep 'target host="' *xml | grep '\*.*\*'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules with targets containing URLs/paths instead of hostnames:"
if grep 'target host="' *xml | egrep '="[^"]*/'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules redirecting to http in to pattern:"
if grep 'to="' *xml | sed 's/^.*to="//' | sed 's/\".*$//' | grep '^http:'
then
   fail=1
else
   echo "(None.)"
fi

echo
echo "-- Rules containing space before to pattern:"
if grep 'to="' *xml | sed 's/^.*to="//' | grep '^ '
then
   fail=1
else
   echo "(None.)"
fi

none=true
echo
echo "-- Rules with syntactically invalid regular expressions:"
for f in *.xml
do
    grep from= "$f" | cut -d\" -f2 | egrep -f - 2>/dev/null || {
        if [ $? -eq 2 ]
        then
            echo "$f"
            fail=1
            none=false
        fi
    }
done
$none && echo "(None.)"

echo
if which xmllint >/dev/null
then
    echo "-- Rules with syntatically invalid XML:"
    none=true
    for rule in *.xml
    do
        xmllint "$rule" >/dev/null 2>&1 || { echo $rule; none=false; fail=1; }
    done
    $none && echo "(None.)"
else
    echo "-- Could not check XML validity because xmllint not found."
fi

echo
echo "-- Rules containing non-ASCII characters (possible homoglyph attacks):"
none=true
for i in *.xml
do
    if egrep '(from|to)=' "$i" | tr -d '[:print:][:space:]' | grep . >/dev/null
    then
        echo "$i contains non-ASCII character(s)."
        none=false
        #fail=1
        echo "(not exiting)"
    fi
done
$none && echo "(None.)"

echo
echo "--- Rules that lack at least one valid <target> tag:"
none=true
for i in *.xml
do
    if ! egrep '<target .*host=".*/>' "$i" >/dev/null
    then
        echo $i
        none=false
        fail=1
    fi
done
$none && echo "(None.)"

echo
echo "--- Duplicated ruleset names:"
if cat *.xml | grep 'ruleset name' | cut -d\" -f2 | sort | uniq -d | grep .
then
    fail=1
else
    echo "(None.)"
fi

echo
echo "--- Duplicated target hosts:"
if cat *.xml | grep '<target host=' | cut -d\" -f2 | sort | uniq -d | grep .
then
    echo "(not exiting)"
else
    echo "(None.)"
fi

exit ${fail-0}