NIST Vulnerability Extractor

#!/bin/bash

# ===================================================================
#   Vulnerability extractor script
#   by Joseph Burger
#   2015 - 3 - 17

#   This shell script takes the url to the vulnerabilities XML file,
#   packaged as a .gz file, from http://static.vbd.nist.gov/downloads.cfm
#   and then downloads it, unzips it, extracts the vulnerability id and score
#   using an awk script, and then further refines the output of that awk script
#   using a sed script. The finished product is a text file with two columns, the
#   first column being the vulnerability id, and the second column having the
#   vulnerability's severity score, or in the case that there's no severity score,
#   "TBD". This output file is saved as "condensedVulns.txt"
# ===================================================================


# ===================================================================
#   Get the xml file.
#   User input method is commented out, in favor of
#   automatically getting it using curl with some egrep
# ===================================================================

echo Please enter the download URL of the vulnerability.xml.gz file
read xmlURL

#xmlURL=`curl https://nvd.nist.gov/download.cfm | \
#egrep feeds/xml.*.gz | tail -n 1 | sed -r "s/.*a href='(.*?)' .*$/\1/"`;

# ===================================================================
#   This command extracts the awk script from this shell script and
#   stores it inside the file "awkScript"
# ===================================================================
tail -n 22 getVulns.sh | tr '\#' ' ' > awkScript

wget $xmlURL && \
gzip -d *.xml.gz && \
awk -f awkScript *.xml | sed -rn '/<entry id/N; s/^.*<entry id*."(.*)".*cvss:score>(.*)<.*$/\1\t\2/w condensedVulns.txt' && \
rm *.xml awkScript
echo $myURL
echo The vulnerabilities are listed in "condensedVulns.txt"

# ===================================================================
#   Store the awk script inside this shell script to save space
# ===================================================================

# This script goes through an xml file containing vulnerabilities
# and extracts the id, as well as the score. If there is no score
# it substitutes "TBD".
#BEGIN{
#   num=0;
#   }
#{
#   if( /entry id/ ) {
#       print $0;
#       arr[num++] = $0
#       while ( ! /cvss:score/ && ! /entry>/ ){
#           getline < FILENAME
#       }
#       if ( /cvss:score/ ) {
#           arr[num] = arr[num] $0
#       } else {
#           arr[num] = arr[num] "\t<cvss:score>TBD</cvss:score>"
#       }
#       while ( ! /entry>/ ) {
#           getline < FILENAME
#       }
#       print arr[num]
#   }
#}
#END{}