# Proxomitron4 URL killfile
# by Paul Rupe
#
# NoAddURL
# NoHash
#
# $Header: E:/RCS/AdList.txt 1.11 2001/12/04 18:46:29 prupe Exp prupe $
#
# Get the latest version of this file from
#   http://www.geocities.com/u82011729/prox/blocklist.html
# Based on
#   Willem Irrwarr's blockfile at
#     http://www.eccentrix.com/computer/wirrwarr/
#   Stephen Martin's Hosts file at
#     http://www.smartin-designs.com/
#   Ryan Farmer's Hosts file at
#     http://www.geocities.com/ryanf86_2000/
#
# Input:
#   Some bounded string representing an absolute URL minus the protocol and
#   :// separator.  Suitable for the URL-Killer header filter.
# Output:
#   On successful match, sets \9 to the keyword (domain, hostname, or path)
#   that caused the match.
#
# This is an attempt to make the ad site filter list more structured and
# easier to maintain.  Looking over Willem's original list you will notice
# that most of the entries fall into one of three categories:
#
# (1) Domains - Entire domains like doubleclick that are blocked with an
#     expression like "([^/]++.|)doubleclick.".
# (2) Hostnames - Common hosts names like ads, adserver, banners that are
#     blocked regardless of the domain.  These are represented by entries
#     such as "ads.*".
# (3) Path components - Paths like /ads/ that are blocked even if the
#     domain and hostnames were ok.  These show up as "*/ads/", for example.
#
# Each of these three types is pushed into a separate list.  This means you
# must configure four lists instead of one:
#   AdList    - AdList.txt
#   AdDomains - AdDomainList.txt
#   AdHosts   - AdHostList.txt
#   AdPaths   - AdPathList.txt
# Of these, only AdList is useful directly in filters.
#
# The advantage to this is that it simplifies adding new entries.  Do you
# want to block an entire domain?  Add its name to the AdDomain list.  No
# need to worry about the proper wildcards to put around it, that part is
# taken care of by AdList.  If suddenly, the boundaries for what constitutes
# a path component change, we merely update the top-level entry in AdList,
# not every single line in AdPaths.
#
# Another big advantage is efficiency.  Since the sublists do not have
# wildcards at the beginning, Proxomitron can use its hashing algorithms on
# them.  Also because of this hierarchy, Proxomitron does not have to keep
# evaluating ([^/]++.|) over and over.  The result is that this version of
# AdList is several times faster than the original.
#
# The disadvantage, of course, is that you now have four lists to install
# rather than one.



# Blocked domains - Entries in this list are single words like doubleclick
# or possibly full domain names like dynamic.dol.ru.  A given entry is
# optionally preceded by a hostname and followed by dot, slash, or
# end-of-string.  The last is so that "http://dynamic.dol.ru" will match even
# without a trailing slash.  Also add colon to take care of URLs with port
# numbers, and quote chars for URLs embedded in scripts.
([^/]++.|)($LST(AdDomains))\9([./:"']|(^?))

# Blocked hostnames - Entries in this list are single words like adserver or
# complete hostnames like ns.netsol.com.  Each entry is checked at the
# beginning of the input string only.  To match it must be followed by a
# boundary: a dot, slash, dash, or end-of-string.  The purpose of this is so
# that an AdHosts entry "ad" won't accidentally block "admin.something.com".
($LST(AdHosts))\9([./\-:"']|(^?))

# Blocked paths - Entries in this list are single words like ads.  Each
# entry is matched against components in the path portion of the URL,
# separated by dot, slash, dash, underscore, question mark, or ampersand.
*[./\-_?&:=]($LST(AdPaths))\9[./\-_?&:="']