# robots.txt file for http://www.sw.org
#
# Note: The default record (User-agent: *) should be the last one in the file.
#       Many bots stop at the first record that matches their user agent,
#       so if you have * listed first, they may never see the specific
#       rules you have for them.
#       Accordingly any rules you have in the default record should also
#       appear in each of the preceding specific bot records that you want
#       to obey such rules.


# The S&W Google Appliance for public sites.
# Allow everything.  Let the Google Mini's exclusion list decide what it will not index.
User-agent: SW-gsa-crawler
Disallow: 

# The S&W Google Appliance for InSite -- exclude all, just in case it gets here unintentionally.
User-agent: SWinsite-gsa-crawler
Disallow: /

# The LinkWalker paid subscription service -- exclude until we have requirements.
#User-agent: linkwalker
#Disallow: /

# Everyone else -- these rules should also exist for any specified public bot.
# Disallow the following:
#   AJAX calls
#   SitePublisher URLs (in case they leak out)
#   Invalid links to the home page that have been discovered by the Google Mini
#   Search results pages returned by the Google Mini
#   Doctor pages that do not use the vanity URL
#   
User-agent: *
Disallow: /ajax-services/
Disallow: /sites/www/
Disallow: /sites/healthcare-professionals/
Disallow: /sites/researchers/
Disallow: /$PAGE_LINK
Disallow: /home_LI
Disallow: /home%
Disallow: /homehome
Disallow: /homeome
Disallow: /homePAGE_LINK
Disallow: /search-results
Disallow: /doctor-detail-pb
Disallow: /ExtProvider