#! /bin/csh -f

# Reads an httpd error log from standard input,
# selects bad accesses to or from local users,
# writes a summary of the form COUNT REFERRER -> BADLINK,
# sorted by user name

grep -v 'Connection ' \
  | grep -v 'Timed out ' \
  | sed \
    -e 's/%7E/~/' \
    -e 's/^\[[^]]*\] \[[^]]*\] //' \
    -e 's/^\[host: [^] ]* referer: \([^]]*\)\]/\1 -> /' \
    -e 's/^\[host: [^ ]*\] /? -> /' \
  | sort \
  | uniq -c \
  | grep -e '\.dcc\.unicamp\.br/[~%]' \
  | sort -t~ +1