#!/bin/bash

# Uncomment to get verbose output
#VERBOSE=1

if test "x$VERBOSE" = x1 ; then set -x; fi

# Constants passed in from configure.ac/CMakeLists
abs_top_srcdir='@abs_top_srcdir@'
abs_top_builddir='@abs_top_builddir@'

# Additional configuration information
. ${abs_top_builddir}/test_common.sh

delta="$1"

# Sanity checks

# 1. This requires that the AWS CLI (command line interface) is installed.
if ! which aws ; then
    echo ">>>> The s3cleanup script requires the \"aws\" command (i.e. the AWS command line interface program)"
    echo ">>>> Try installing \"awscli\" package with apt or equivalent."
    exit 0
fi

# 2. Make sure S3TESTSUBTREE is defined
if test "x$S3TESTSUBTREE" = x ; then
    echo ">>>> The s3cleanup script requires that S3TESTSUBTREE is defined."
    exit 1
fi

# 3. Make sure delta is defined
if test "x$delta" = x ; then
    echo ">>>> No delta argument provided"
    echo ">>>> Usage: s3gc <delta>"
    echo ">>>>        where <delta> is number of days prior to today to begin cleanup"
    exit 1
fi


# This script takes a delta (in days) as an argument.
# It then removes from the Unidata S3 bucket those keys
# that are older than (current_date - delta).

# Compute current_date - delta

# current date
current=`date +%s`
# convert delta to seconds
deltasec=$((delta*24*60*60))
# Compute cleanup point
lastdate=$((current-deltasec))

rm -f s3gc.json

# Get complete set of keys in ${S3TESTSUBTREE} prefix
if ! aws s3api list-objects-v2 --bucket ${S3TESTBUCKET} --prefix "${S3TESTSUBTREE}" | grep -F '"Key":' >s3gc.keys ; then
    echo "No keys found"
    rm -f s3gc.json
    exit 0
fi
aws s3api list-objects-v2 --bucket ${S3TESTBUCKET} --prefix "${S3TESTSUBTREE}" | grep -F '"Key":' >s3gc.keys
while read -r line; do
  KEY0=`echo "$line" | sed -e 's|[^"]*"Key":[^"]*"\([^"]*\)".*|\1|'`
  # Strip off any leading '/'
  KEY=`echo "$KEY0" | sed -e 's|^[/]*\(.*\)|\1|'`
  # Ignore keys that do not start with ${S3TESTSUBTREE}
  PREFIX=`echo "$KEY" | sed -e 's|\([^/]*\)/.*|\1|'`
  if test "x$PREFIX" = "x$S3TESTSUBTREE" ; then
      ALLKEYS="$ALLKEYS $KEY"
  fi
done < s3gc.keys

# Look at each key and see if it is less than lastdate.
# If so, then record that key

# Capture the keys with old uids to delete
unset MATCHKEYS
for key in $ALLKEYS ; do
    case "$key" in
        "$S3TESTSUBTREE/testset_"*)
             # Capture the uid for this key         
             s3uid=`echo $key | sed -e "s|$S3TESTSUBTREE/testset_\([0-9][0-9]*\)/.*|\1|"`
             # check that we got a uid       
             if test "x$s3uid" != x ; then
                # Test age of the uid
                if test $((s3uid < lastdate)) = 1; then
                    MATCHKEYS="${MATCHKEYS} $key"
                fi
             else
                 if test "x$VERBOSE" = x1 ; then echo "Excluding \"$key\""; fi
             fi
	     ;;
        *) if test "x$VERBOSE" = x1; then echo "Ignoring \"$key\""; fi ;;
    esac
done

# We can delete at most 1000 objects at a time, so divide into sets of size 500
REM="$MATCHKEYS"
while test "x$REM" != x ; do
    K500=`echo "$REM" | cut -d' ' -f 1-500`
    REM=`echo "$REM" | cut -d' ' -f 501-`
    unset DELLIST
    MATCH=0
    FIRST=1
    DELLIST="{\"Objects\":["
    for key in $K500 ; do
        if test $FIRST = 0 ; then DELLIST="${DELLIST},"; fi
        DELLIST="${DELLIST}
{\"Key\":\"$key\"}"
        FIRST=0
        MATCH=1
    done
    DELLIST="${DELLIST}],\"Quiet\":false}"
    rm -f s3gc.json
    if test "x$MATCH" = x1 ;then
	rm -f s3gc.json
        echo "$DELLIST" > s3gc.json
        aws s3api delete-objects --bucket ${S3TESTBUCKET} --delete "file://s3gc.json"
    fi
done
rm -f s3gc.json