#Strip emails from mediauk via some horrific re hacks + direct parsing of HTML.

filthy_html_email_stripper.sh

#Strip emails from mediauk via some horrific re hacks + direct parsing of HTML.
#  Their site breaks Beautiful Soup :(

#B wants all the email addresses in this linked to page
wget -O - "http://www.mediauk.com/tags/local" | sed 's/>/\n/g' | grep "href" | cut -f2 -d\'  | grep "^\/" > urls.list
#I think I manually deleted the false ones in the top + bottom of urls.list

for URL in ` cat urls.list `
do
  echo -n "${URL}    "
  email=` wget -O - "http://www.mediauk.com${URL}" | sed 's/>/\n/g' | grep "og:email" | cut -f3 -d\  | cut -f2 -d\"  `
  echo "${email}" #ensures end of line...

done