Is it possible to block website accessing based on HTML's "header words" by squid proxy on RHEL 5 ?
- If possible please tell as detail.
What do you mean by 'header words'? Do you mean something win the HTML body or the HTTP headers?
Anyway if you want to block something in the body of the web page for no cost you probably should be looking at DansGuardian.
You can use Squids ACLs (Access Control Lists) to limit Access to certain pages - based on different criteria. Sounds like you are looking for a "rep_header" check.
"rep_header: regular expression pattern matching on a reply header content..."
See http://wiki.squid-cache.org/SquidFaq/SquidAcl#ACL_elements for more options. But setting up ACLs using regex can be quite tricky. Maybe you can google around.