submission-report

#/bin/bash

# author Renzo Kottmann


declare -r DEBUG='true'


function log::debug () {
  test -z "${DEBUG}" && { echo "ERROR: DEBUG variable not set"; exit 100; }
  test -n "${DEBUG}" && echo "DEBUG ${1}" 1>&2;
}


function log::error () {
   test -z "${DEBUG}" && { echo "ERROR: DEBUG variable not set"; exit 100; }
   test -n "${DEBUG}" && echo "ERROR ${1}" 1>&2;
}


function check_empty_var() {
    if [[ ! ${1} ]]; then
        log::error "Variable is empty";
        return 100
    fi
    return 0
}

function create_usage() {
    cat <<EOF

Usage: ${0}

Options:

  -s, --sequencing-center  Provide a sequencing center name e.g. LGC_GENOMICS

  -c, --campaign-name      Provide a campaign name e.g. OSD2014 or MYOSD2016

  -d, --data-set           Provide the data set name e.g. 18S, 16S or shotgun

  -p, --processing-status  Provide a th (pre-) processing status of sequences e.g. RAW or WORKABLE

  --primer                 Provide a primer pair name used for sequencing e.g. alma_alma or NONE in case of shotgun

EOF
}

function main() {

  check_empty_var "${1}" || { log::debug 'Provide options';  create_usage; }

  # expecting last argument to be the directory from where to gather the report
  while [[ $# -gt 0 ]]; do

     case ${1} in

        -s|--sequencing-center)
            shift;
            check_empty_var "${1}" || {
                log::error "Provide a sequencing center name e.g. LGC_GENOMICS";
                exit 66
            }

            local -r SEQUENCING_CENTER="${1}"

            shift;
            ;;
        -c|--campaign-name)
            shift;
            check_empty_var "${1}" || {
                log::error "Provide a campaign name e.g. OSD or MYOSD";
                exit 66
            }

            local -r CAMPAIGN_NAME="${1}"

            shift;
            ;;
        -d|--data-set)
            shift;
            check_empty_var "${1}" || {
                log::error "Provide the sequencing data category e.g. 16S or shotgun";
                exit 66
            }

            local -r DATA_CAT="${1^^}"

            shift;
            ;;
        -p|--processing-status)
            shift;
            check_empty_var "${1}" || {
                log::error "Provide a th (pre-) processing status of sequences e.g. RAW or WORKABLE";
                exit 66
            }

            local -r PROCESSING_STATUS="${1,,}"

            shift;
            ;;
        --primer)
            shift;
            check_empty_var "${1}" || {
                log::error "Provide a the primer pair used for sequencing e.g. alma_alma or NONE in case of shotgung";
                exit 66
            }

            local -r PRIMER_PAIR="${1,,}"

            shift;
            ;;
        -h|--help)
            create_usage ;
            exit 0
            ;;
        *)
            log::error "Unknown option  $1"
            log::debug  "Option  $1 unknown"
            declare -gr dir="${1}"
            # now checking last argument
            if [[ -d ${1} ]]; then
              echo "dir=${dir} exists. Can proceed..."
            else
              log::error "dir=${dir} does not exist"
              exit 1
            fi
            shift;
            ;;
    esac
  done

  check_empty_var "${SEQUENCING_CENTER}" || {
    log::error "SEQUENCING_CENTER missing but required"
    exit 66
  }
  check_empty_var "${CAMPAIGN_NAME}" || {
    log::error "CAMPAIGN_NAME missing but required"
    exit 66
  }
  check_empty_var "${DATA_CAT}" || {
    log::error "DATA SET name missing but required"
    exit 66
  }
  check_empty_var "${PROCESSING_STATUS}" || {
    log::error "PROCESSING_STATUS missing but required"
    exit 66
  }
  check_empty_var "${PRIMER_PAIR}" || {
    log::error "PRIMER_PAIR missing but required"
    exit 66
  }




  log::debug "
SEQUENCING_CENTER=${SEQUENCING_CENTER}
CAMPAIGN_NAME=${CAMPAIGN_NAME}
DATA SET=${DATA_CAT}
PROCESSING_STATUS=${PROCESSING_STATUS}
PIMER PAIR=${PRIMER_PAIR}
"


  local -r CUR_DATE=$(date --rfc-3339=date)
  local md5_report="submission_md5_${CAMPAIGN_NAME}_${PROCESSING_STATUS}_${DATA_CAT}_report_${CUR_DATE}.txt"
  local submission_report="submission_${CAMPAIGN_NAME}_${PROCESSING_STATUS}_${DATA_CAT}_files_report_${CUR_DATE}.csv"


  rm --force --verbose "${submission_report}"
  log::debug "dir=${dir}"

  shopt -s extglob

  cat "${dir}/"*.md5 > ${md5_report}

  cat ${md5_report}  | \
    while read md5 filename; do
      # if md5 is from binary file there is * at beginning, therefore replace with nothing in case
      filename=${filename/#\*/}
      filename=$(basename "${filename}")
      pp_fullpath=$(readlink -f "${dir}/${filename}")
      log::debug "
        filename=${filename}
        dir=${dir}
        fullpath=${pp_fullpath}
      "
      echo "${CAMPAIGN_NAME},${SEQUENCING_CENTER},${DATA_CAT},${PROCESSING_STATUS},${PRIMER_PAIR},${md5},${filename},${pp_fullpath}" >> ${submission_report} ;
    done
}

log::debug 'Hi Submission Reporter'

main "$@"