##### strip_egroups_ad.rc
#
# Resource file for procmail. Run with:
#   INCLUDERC=yourpath/strip_egroups_ad.rc
# in your $HOME/.procmailrc
#
# Removes the first ad in eGroups list emails.
# First and last line of ad are matched by $start and $end, and some string
# inside the ad by $admatch. The whole ad is replaced by $note.
#
# The usual headache with bare-bones Unix-rubbish: the sed solution never(!)
# works under Solaris 2.7 because -e handles neither newlines nor nested
# '{ }'-lists. Solaris awk is also too dumb - nawk is required.
# Needless to say, the GNU tools never have a problem. Long live the Penguin!
#
# The latest version is always available from:
# http://volker.orcon.net.nz/soft/procmail/strip_egroups_ad.rc
#
# In the public domain.
# Volker Kuhlmann <v.kuhlmann@elec.canterbury.ac.nz>
#   31 Aug 2000; 12 Jan; 2 Feb; 4, 29 Mar 2001; 10, 11, 14 Apr 2002
#

:0
* ^Delivered-To:.*@(yahoogroups|egroups)\.com
* ^Mailing-List:.*@(yahoogroups|egroups)\.com

# With awk (change to nawk, gawk etc. if necessary, e.g. solaris awk is bad):
# 10Apr02: Arrrgh, with nawk/gawk one has to use "\\." to get a literal "."!
# The replacement string can now be empty, i.e. note=''.
#
{
  # detect version of awk to use, in order: gawk, nawk, awk
  :0 i
  * AWK ?? ^^^^
  AWK=| sh -c 'for awk in gawk nawk awk; do \
  		($awk </dev/null "") 2>/dev/null >&2 \
  		  && { echo $awk; break; } done'
  # or set variable here and comment out the 5 lines above:
  #AWK=gawk
  
  end='...-----------------.*-----------------([-~>=_|e]*[~>=_|e]|[^-].*-)$'
  start="^$end"
  admatch='http://.*\\.(yahoo|egroups)\\.com/.*\\/'
  note='[obnoxious Yahoo/eGroups ad removed]\n'
  :0 fbw
  * ! AWK ?? ^^^^
  | $AWK "\
      BEGIN { ad=0; done=0 }\
      done { print; next }\
      ad && \$0 ~ \"$end\" { \
        ad=0; \
	if (match(text,\"$admatch\")) {\
	  printf \"%s\", \"$note\"; done=1\
	} else {\
	  print text \$0; text=\"\" }\
	next\
      }\
      \$0 ~ \"$start\" { ad=1 }\
      ad { text=text \$0 \"\n\" }\
      !ad { print }\
      "
}

# With sed:
#
# Write $start and $end to match the whole line, but do not(!) anchor $end at
# the start of the line using "^".
# The /$admatch/! condition is necessary to remove the first ad line, in case
# $end also matches the start line.
# The conditions should be reasonably broad, and still catch if egroups changes
# some characters in the lines.
#
# Adopted from the sed FAQ:
#   :t
#   /BLOCK_TOP/,/BLOCK_END/ {
#     /BLOCK_END/! { N; b t; }
#     /regex/s/^.*BLOCK_END//
#   }
#   Suppose the beginning of the block is indicated by 'BLOCK_TOP' and
#   the end of the block is indicated by 'BLOCK_END'. If the expression
#   'regex' appears anywhere within the block, the entire block should
#   be deleted.
# The most difficult part was to get the quoting right for procmail...
# Note: this doesn't trigger on old email with egroups.com URL in the ad any
# more.
#
#{
#  end='...-\{20,\}.*-\{20,\}[-~>=_|e]*$'
#  start="^$end"
#  admatch='http:\/\/.*\.yahoo\.com\/.*\/'
#  note='\[obnoxious Yahoo\/eGroups ad removed\]'
#  :0 fbw
#  | sed \
#    -e ':t' \
#    -e "/$start/,/^$end/ {        \
#          /^.*$end/! { N; b t; }; \
#	  /$admatch/! { N; b t; }; \
#	  /$admatch/ {             \
#	    s/^.*$end/$note/ ;     \
#	    :tt;                   \
#	    n; b tt;               \
#          };                      \
#	}"
#}

##### EOF strip_egroups_ad.rc