obfuscate.sh

   1 #!/bin/bash
   2
   3 # obfuscate.sh
   4 # obfuscate email addresses in XML/HTML
   5 # Script written (and slight perl modification) by Archaic <archaic AT linuxfromscratch D0T org>
   6 # Modified from "sed -i" to old style "sed -e" by Manuel Canales <manuel AT linuxfromscratch D0T org>
   7 # to prevent hangs on very long files, like nonckunked books.
   8 # Original Perl expression by Anderson Lizardo <lizardo AT linuxfromscratch D0T org>
   9 # Released under the GNU General Public License
  10 #
  11 # This script currently only seeks out mailto: addresses. If those same
  12 # addresses also appear in plaintext, we need to obfuscate those as well.
  13 #
  14 # This script was made for a very specific purpose so I was a bit lazy in
  15 # writing the regex's.
  16 #
  17 # Please send comments, enhancements, etc. to the above address
  18
  19 #set -e  # Bail on all errors
  20
  21 # First, ensure that we are given a file to process
  22 # if [ $# -lt 1 ]; then
  23 #   echo -e "\nYou must provide an input file."
  24 #   exit 1
  25 # fi
  26
  27 # Nothing like a backup plan!
  28 #cp "$1" "$1".bak
  29
  30 for i in `grep -o '"mailto:.*@.*"' ${1} |sed -e 's|^"mailto:||' -e 's|"$||'`; do
  31   link=`echo $i | perl -pe 's/[^\n]/"\\\&#".ord($&)."\;"/ge'`
  32   plaintext=`echo $i | sed -e 's|@| AT |' -e 's|\.| D0T |g'`
  33   cp ${1}{,.tmp}
  34   sed -e "s|mailto:$i|mailto:$link|" \
  35       -e "s|$i|$plaintext|" ${1}.tmp > ${1}
  36   rm ${1}.tmp
  37 done
  38 #rm $FILE.tmp
  39 #exit 0