#!/bin/bash

# @project      The CERN Tape Archive (CTA)
# @copyright    Copyright © 2022 CERN
# @license      This program is free software, distributed under the terms of the GNU General Public
#               Licence version 3 (GPL Version 3), copied verbatim in the file "COPYING". You can
#               redistribute it and/or modify it under the terms of the GPL Version 3, or (at your
#               option) any later version.
#
#               This program is distributed in the hope that it will be useful, but WITHOUT ANY
#               WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
#               PARTICULAR PURPOSE. See the GNU General Public License for more details.
#
#               In applying this licence, CERN does not waive the privileges and immunities
#               granted to it by virtue of its status as an Intergovernmental Organization or
#               submit itself to any jurisdiction.



echo "$(date +%s): Creating test dir in eos: ${EOS_DIR}"
# uuid should be unique no need to remove dir before...
# XrdSecPROTOCOL=sss eos -r 0 0 root://${EOSINSTANCE} rm -Fr ${EOS_DIR}


eos root://${EOSINSTANCE} mkdir -p ${EOS_DIR} || die "Cannot create directory ${EOS_DIR} in eos instance ${EOSINSTANCE}."

echo
echo "Listing the EOS extended attributes of ${EOS_DIR}"
eos root://${EOSINSTANCE} attr ls ${EOS_DIR}
echo


# As we are skipping n bytes per file we need a bit more than the file size to accomodate dd to read ${FILE_KB_SIZE} skipping the n first bytes
dd if=/dev/urandom of=/tmp/testfile bs=1k count=$((${FILE_KB_SIZE} + ${NB_FILES}*${NB_DIRS}/1024 + 1)) || exit 1


db_begin_transaction

# not more than 100k files per directory so that we can rm and find as a standard user
for ((subdir=0; subdir < ${NB_DIRS}; subdir++)); do
  for((slot=1; slot <= ${NB_PROCS}; slot++)); do
    touch "slot${slot}"
  done

  eos root://${EOSINSTANCE} mkdir -p ${EOS_DIR}/${subdir} || die "Cannot create directory ${EOS_DIR}/{subdir} in eos instance ${EOSINSTANCE}."

  echo -n "Copying files to ${EOS_DIR}/${subdir} using ${NB_PROCS} processes..."
  TEST_FILE_NAME_SUBDIR=${TEST_FILE_NAME_BASE}$(printf %.2d ${subdir}) # this is the target filename of xrdcp processes just need to add the filenumber in each directory
  
  file_creation="dd if=/tmp/testfile bs=1k 2>/dev/null | (dd bs=$((${subdir}*${NB_FILES})) count=1 of=/dev/null 2>/dev/null; dd bs=TEST_FILE_NUM count=1 of=/dev/null 2>/dev/null; dd bs=1k count=${FILE_KB_SIZE} 2>/dev/null) "
  
  xrdcp_call="XRD_LOGLEVEL=Dump xrdcp - root://${EOSINSTANCE}/${EOS_DIR}/${subdir}/${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM 2>${ERROR_DIR}/${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM"
  
  xrdcp_succes=" rm ${ERROR_DIR}/${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM  && echo ${subdir}/${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM  >> slot\"{%}\""

  xrdcp_error="ERROR with xrootd transfer for file ${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM, full logs in ${ERROR_DIR}/${TEST_FILE_NAME_SUBDIR}TEST_FILE_NUM"

  command_str="${file_creation} | ${xrdcp_call} && ${xrdcp_succes} || ${xrdcp_error}"
   
  for ((i=0;i<${NB_FILES};i++)); do
    echo $(printf %.6d $i)
  done | parallel --max-procs=${NB_PROCS} -iTEST_FILE_NUM bash -c "true && $command_str"
  #TODO: Figure out why we need that 'true &&' in parallel and not for xargs.

  # Initialize db
  for((i=1; i <= ${NB_PROCS}; i++)); do
    cat slot${i} | xargs -iFILE bash -c "db_insert FILE"
    rm -f slot${i}
  done

  echo Done.
done
db_commit_transaction
db_begin_transaction
if [ "0" != "$(ls ${ERROR_DIR} 2> /dev/null | wc -l)" ]; then
  # there were some xrdcp errors
  echo "Several xrdcp errors occured during archival!"
  echo "Please check client pod logs in artifacts"
  mv ${ERROR_DIR}/* ${LOGDIR}/xrd_errors/
fi

COPIED=0
COPIED_EMPTY=0
for ((subdir=0; subdir < ${NB_DIRS}; subdir++)); do
  COPIED=$(( ${COPIED} + $(eos root://${EOSINSTANCE} find -f ${EOS_DIR}/${subdir} | wc -l) ))
  COPIED_EMPTY=$(( ${COPIED_EMPTY} + $(eos root://${EOSINSTANCE} find -0 ${EOS_DIR}/${subdir} | wc -l) ))
done

# Only not empty files are archived by CTA
TO_BE_ARCHIVED=$((${COPIED} - ${COPIED_EMPTY}))

ARCHIVED=0
echo "$(date +%s): Waiting for files to be on tape:"
SECONDS_PASSED=0
WAIT_FOR_ARCHIVED_FILE_TIMEOUT=$((40+${NB_FILES}/5))
TMPFILE=$(mktemp)
while test ${TO_BE_ARCHIVED} != ${ARCHIVED}; do
  echo "$(date +%s): Waiting for files to be archived to tape: Seconds passed = ${SECONDS_PASSED}"
  sleep 3
  let SECONDS_PASSED=SECONDS_PASSED+1

  if test ${SECONDS_PASSED} == ${WAIT_FOR_ARCHIVED_FILE_TIMEOUT}; then
    echo "$(date +%s): Timed out after ${WAIT_FOR_ARCHIVED_FILE_TIMEOUT} seconds waiting for file to be archived to tape"
    break
  fi

  ARCHIVED=0
  for ((subdir=0; subdir < ${NB_DIRS}; subdir++)); do
    eos root://${EOSINSTANCE} ls -y ${EOS_DIR}/${subdir} | grep '^d0::t1' | awk '{ print $10 }' | sed "s|^|${subdir}/|" > ${TMPFILE}
    db_update_from_file ${TMPFILE} archived 0
    ARCHIVED=$(( ${ARCHIVED} + $(cat ${TMPFILE} | wc -l) ))
    sleep 1 # do not hammer eos too hard
  done

  echo "${ARCHIVED}/${TO_BE_ARCHIVED} archived"

  NB_TAPE_NOT_FULL=`admin_cta --json ta ls --all | jq "[.[] | select(.full == false)] | length"`
  if [[ ${NB_TAPE_NOT_FULL} == 0 ]]
  then
    echo "$(date +%s): All tapes are full, exiting archiving loop"
    break
  fi
done
rm -f ${TMPFILE}
db_commit_transaction


echo "###"
echo "${ARCHIVED}/${TO_BE_ARCHIVED} archived"
echo "###"

echo "Archiving done."
echo "###"
echo "${TAPEONLY}/${ARCHIVED} on tape only"
echo "###"
echo "Sleeping 10 seconds to allow MGM-FST communication to settle after disk copy deletion."
sleep 10
echo "###"
db_info "*" 10