#!/bin/sh
#
# sakai-files-dedup - deduplicate Sakai's uploaded files
#
# $Id$

# designed to be called from /etc/cron.daily/

# for testing, use e.g.
#
#  MIN_SIZE=1M sakai-files-dedup -n
#
# or
#
#  MIN_SIZE=1 sakai-files-dedup -na
#
# the flags are passed to ln-r

# defaults overridable in environment variables:
: ${DEDUP_USER:=tomcat8s11}
: ${MIN_SIZE:=100M}
#: ${RECIPIENT:=sakai-remove-duplicates}
# set $RECIPIENT to send the results in email to that recipient
# by default, no email is sent and the results go to stdout

PATH=/usr/bin:/bin

Die() { echo "${0}: fatal error: $@"; exit 1; }

DieIfMissing()
{
   type "$1" >/dev/null 2>/dev/null ||
   Die command not found: "$1"
}

DieIfNoUser()
{
   getent passwd "$1" >/dev/null ||
   Die user not found: "$1"
}

# execute as user $DEDUP_USER

DieIfNoUser "$DEDUP_USER"

# use sudo if DEDUP_USER is someone else
[ $DEDUP_USER = "`whoami`" ] ||
  exec sudo -u "$DEDUP_USER" -g "$DEDUP_USER" MIN_SIZE="$MIN_SIZE" RECIPIENT="$RECIPIENT" "$0" "$@"
  # $@ is just for passing flags to ln-r, such as -n!

# executing as user $DEDUP_USER

HOME=`getent passwd "$DEDUP_USER" | awk -F: '{print $6}'`
D="$HOME"/tomcat8/sakai/files

DieIfMissing realpath

ME=`realpath "$0"`
LN_R=`dirname "$ME"`/ln-r

[ -x "$LN_R" ] || Die command not found: ln-r

cd "$D" || Die cannot cd to "$D"

DedupAndReport()
# any arguments are assumed to be flags; they are passed to ln-r;
# this is useful for testing with -n and -a
{
  # assume no spaces etc. in filenames
  find 20?? -type f -size +"$MIN_SIZE" |
    "$LN_R" -Iv "$@" 2>&1 |
    awk '{print $NF}' |
    perl -lne '
      BEGIN { $c = 0; $b = 0 };
      ++$c;
      $b += (stat($_))[7];
      END { print "deduplicated $c files, $b bytes" if $c > 0 }' |
    perl -wpe '1 while s/(\d+)(\d\d\d)/$1,$2/;'
  # adds thousands separators
}

if [ "$RECIPIENT" = "" ]
then
  DedupAndReport "$@"
  # may be empty
else
  DEDUP=`DedupAndReport "$@"`

  if [ "$DEDUP" != "" ]
  then
    # we have output
    DieIfMissing mail
    echo "$0: $DEDUP" |
      mail -s "removing duplicate files: $DEDUP" "$RECIPIENT"
  fi
fi