#!/bin/sh

# Usage:
#
#   read mx_token
#   mx_token=$mx_token sudo -E synapse-maint

set -e -u

case "${1:-}" in
  srun)
    exec systemd-run --collect --uid=matrix-synapse \
      --unit=synapse-maintainance -E mx_token \
      --nice=4 -pCPUSchedulingPolicy=batch -pIOSchedulingClass=idle $0
    ;;

  log)
    shift
    if test $# -eq 0
    then
        set -- -b --since=-10days
    fi

    exec journalctl -u synapse-maintainance "$@"
    ;;
esac

PATH=/home/joerg/bin:$PATH
db=matrix
conn_str="host=/run/postgresql user=$(id -u -n) dbname=$db application_name=state_compress"
: ${synapse_admin_url:=http://localhost:8008/_synapse/admin/v1}

export RUST_BACKTRACE=short LC_ALL=C.UTF-8

if which time >/dev/null
then
    export TIME='%E elapsed %M rss'
    time=time
else
    time=
fi

psql() {
    command psql -X -b "$@" "dbname=$db application_name=synapse-maint"
}

psql_t() {
    ${time:-command} psql -X -b "$@" "dbname=$db application_name=synapse-maint"
}

if test -n "$mx_token"
then
    echo 'Removing orphaned rooms'
    echo '======================='
    echo

    for room in $(
      curl -s --header "Authorization: Bearer $mx_token" "$synapse_admin_url/rooms?limit=800" \
          |jq -r '.rooms[] |select(.joined_local_members == 0) | .room_id'
    )
    do
        echo -n "Purging $room: "
        $time curl -s -X DELETE -H "Authorization: Bearer $mx_token" \
          -H 'Accept: application/json' -H "Content-Type: application/json" \
          -d '{"purge": true, "message": "Sorry - kicking you out to clean up the database"}' \
          "$synapse_admin_url/rooms/$room"
    done
fi

# see _purge_room_txn in synapse/storage/databases/main/purge_events.py (v1.98.0)

echo 'Removing data referencing unknown rooms'
echo '======================================='
echo

for tbl in event_auth event_edges event_json events device_lists_changes_in_room \
    state_groups state_groups_state state_events
do
    printf '%s: ' "$tbl"
    psql_t -c "DELETE FROM $tbl WHERE room_id NOT IN (SELECT room_id FROM rooms);"
done

# TODO this takes too much time
# echo 'Removing data referencing unknown events'
# echo '========================================'
# echo

# for tbl in event_json state_groups state_groups_state
# do
#     printf '%s: ' "$tbl"
#     psql_t -c "DELETE FROM $tbl WHERE event_id NOT IN (SELECT event_id FROM events);"
# done

# cargo install --git='https://github.com/matrix-org/rust-synapse-compress-state' synapse_auto_compressor
if which synapse_compress_state >/dev/null
then
    echo 'Compressing state'
    echo '================='
    echo

    cd "$(mktemp -d)"
    echo "Writing SQL files to $PWD"

    # SELECT set_config('autovacuum', 'off'', false); SELECT pg_reload_conf(); ?

    echo 'Disabling autovacuum on state_groups_state'
    psql -q -c 'alter table state_groups_state set (autovacuum_enabled = false);'

    for room in $(
      psql -t -A \
          -c 'SELECT room_id FROM state_groups GROUP BY room_id ORDER BY count(*)'
    )
    do
        echo '--------------------'
        echo

        sql_file=state-compress-$room.sql
        $time synapse_compress_state -t -p "$conn_str" -o $sql_file -r $room -m 1000
        if test -s $sql_file
        then
            psql_t -q -c '\set ON_ERROR_STOP on' -f $sql_file
        else
            rm $sql_file
        fi
    done

    echo 'Enabling autovacuum on state_groups_state'
    psql -q -c 'alter table state_groups_state set (autovacuum_enabled = true);'

    echo 'Running VACUUM and ANALYZE for state_groups_state ...'
    psql_t -q -c 'VACUUM FULL ANALYZE state_groups_state'

    echo "All SQL scripts are in $PWD"
fi

psql_t -q -c "REINDEX DATABASE CONCURRENTLY $db"