#! /bin/bash
#------------------------------------------------------------------------------
#$Author: antanas $
#$Date: 2024-08-05 17:01:54 +0300 (Mon, 05 Aug 2024) $
#$Revision: 10207 $
#$URL: svn+ssh://www.crystallography.net/home/coder/svn-repositories/cod-tools/tags/v3.11.0/scripts/cif_cod_deposit $
#------------------------------------------------------------------------------
#*
#* Deposit CIFs into COD database using CGI deposition interface.
#*
#* USAGE:
#*    $0 --options structure1.cif structure*.cif
#**

TMP_DIR="${TMPDIR}"

set -ue
## set -x

BASENAME="$(basename "$0")"

script() { echo "$*"; }
setvar() { eval $1="'$3'"; }

setvar INPUT_CONFIG    = "${HOME:-.}/.cod_deposit.cfg"
setvar SERVER          = "http://test.crystallography.net"
setvar DEPOSIT_PATH    = "cgi-bin/cif-deposit.pl"
setvar DEPOSIT_URL     = "${SERVER}/${DEPOSIT_PATH}"
setvar DEPOSITION_TYPE = "published"

setvar CIF_FILES        = ''
setvar HKL_FILES        = ''
setvar SEPARATE_OUT_DIR = ''
setvar SEPARATOR        = ' '
setvar READ_STDIN       = false
setvar LOG_MESSAGE      = ''

setvar DEBUG    = false
setvar PROGRESS = false
setvar REPLACE  = false
setvar RELEASE  = false
setvar OUTPUT_MODE  = "html"
setvar STRICT     = true
setvar TIMESTAMPS = true
setvar USE_WIPE   = true

setvar REQUIRE_ONLY_DOI_ARG = ''

# CIF_HEADER:
#
# curl http://www.crystallography.net/cod/cgi-bin/cif-deposit.pl -F
# header=crystaleye <other parameters>
#
# Alternative headers can also be used: 'common', 'amcsd', 'iucr'.
#
setvar CIF_HEADER = ''

# Executables:

curl=curl

#* OPTIONS:
#*
#*   -c, --config config-file-with-password.cfg
#*                     Specify which file holds depositor user name and
#*                     password. Default: "~/.cod_deposit.cfg".
#*
#*   -h, --host  test.crystallography.net
#*                     Specify the host to which the structure should be
#*                     deposited to.
#*
#*   -p, --path  cgi-bin/cif-deposit.pl
#*                     Specify script path within the host to use for structure
#*                     upload.
#*
#*   -u, --url  http://test.crystallography.net/cgi-bin/cif-deposit.pl
#*                     Specify the full URL to use for structure uploading.
#*                     Default: combine --host and --path values using
#*                     the "http://" protocol and "/".
#*
#*   -d, --deposition-type published
#*                     Specify the deposition type under which the structure
#*                     should be deposited. Accepted values:
#*                     ["published", "personal", "prepublication"].
#*                     Default: "published".
#*
#*   -s, --separate-outputs folder-to-put-deposition-outputs/
#*                     Specify a folder to put deposition output for each
#*                     deposited file.
#*
#*   -r, --read-stdin
#*                     Read a list of CIF and Fobs file names from the standard
#*                     input in format
#*                     <CIF file> [<Fobs file>]*
#*                     for example:
#*                     Ice.cif
#*                     bi5014sup1.cif all.hkl
#*                     bi5005sup1.cif bi50051sup2.hkl bi50052sup3.hkl
#*
#*   --cif "input1.cif input*.cif"
#*                     Specify a list of CIF files to be deposited.
#*   --add-cif "input1.cif input*.cif"
#*                     Specify additional CIF files to be deposited.
#*   --clear-cif
#*                     Clear the list of CIF files to be deposited.
#*
#*   --hkl, --fobs
#*                     Specify a list of Fobs files to be deposited.
#*   --add-hkl, --add-fobs
#*                     Specify additional Fobs files to be deposited.
#*   --clear-hkl, --clear-fobs
#*                     Clear the list of Fobs files to be deposited.
#*
#*   --replace
#*                     Specify that the given files are replacements for
#*                     previously deposited data.
#*
#*   --release-data
#*                     Specify that given files were previously deposited as
#*                     prepublication material and should be now released into
#*                     the public domain.
#*
#*   --log-message, --log-msg
#*                     Specify the log message for redeposited structures.
#*
#*   --script
#*                     Do not deposit, print the commands to be executed
#*                     on deposition instead.
#*   --run
#*                     Enable the execution of deposition commands instead
#*                     of simply printing them. Default.
#*
#*   --strict,
#*   --no-ignore-unpaired-datablocks, --dont-ignore-unpaired-datablocks
#*                     Stop the deposition if any CIF data block cannot be
#*                     paired to any Fobs data blocks (provided that any
#*                     unpaired data blocks exist). Default.
#*   --silent, --ignore-unpaired-datablocks
#*                     Ignore unpaired CIF and Fobs data blocks.
#*
#*   --require-only-doi
#*                     Do not require any other bibliographic information to be
#*                     present if CIF data block contains resolvable DOI.
#*   --require-full-bibliography
#*                     Require all bibliographic information to be present in
#*                     submitted CIF file. Default.
#*
#*   --use-wipe
#*                     Use the 'wipe' command to remove temporary password
#*                     files. Revert to using 'rm' if 'wipe' is not available.
#*                     Default.
#*   --no-use-wipe, --dont-use-wipe, --use-rm
#*                     Use the 'rm' command to remove temporary password files.
#*                     
#*
#*   --common-cif-header
#*                     Use the standard COD CIF header. Default.
#*   --iucr-cif-header
#*                     Use the COD CIF header agreed with IUCr.
#*   --amcsd-cif-header
#*                     Use the COD CIF header agreed with R.D.
#*   --crystaleye-cif-header
#*                     Use the COD CIF header agreed with P.M.-R.
#*
#*   --print-timestamps
#*                     Print timestamps to STDOUT at the start and end of
#*                     each CIF file deposition. Default.
#*   --no-print-timestamps, --dont-print-timestamps
#*                     Do not print timestamps to STDOUT at the start and end
#*                     of each CIF file deposition.
#*
#*   -P, --progress
#*                     Ask cif-deposit.pl to report progress of the deposition.
#*   -P-, --no-progress
#*                     Silence cif-deposit.pl deposition progress report messages.
#*                     Default.
#*
#*   --output-mode html
#*                     Specify the output mode. Accepted values:
#*                     ["html", "stdout"]. Default: "html".
#*
#*   -D, --debug
#*                     Request output of the progress information from
#*                     cif-deposit.pl to the HTTP server log.
#*   -D-, --no-debug
#*                     Disable the output of the progress information from
#*                     cif-deposit.pl to the HTTP server log. Default.
#*
#*   --help, --usage
#*                     Output a short help message (this message) and exit.
#*   --version
#*                     Output version information and exit.
#**
while [ $# -gt 0 ]
do
  case $1 in
      --cif)
          CIF_FILES="$2"
          shift
          ;;
      --add-cif)
          CIF_FILES="${CIF_FILES} $2"
          shift
          ;;
      --clear-cif)
          CIF_FILES=""
          ;;
      --hkl|--fobs)
          HKL_FILES="$2"
          shift
          ;;
      --add-hkl|--add-fobs)
          HKL_FILES="${HKL_FILES} $2"
          shift
          ;;
      --clear-hkl|--clear-fobs)
          HKL_FILES=""
          ;;
      -c|--config|--confi|--conf|--con)
          INPUT_CONFIG="$2"
          DEPOSIT_URL="$(perl -lne "print \$1 if /^server_url=(.*)/" \
                         < "${INPUT_CONFIG}")"
          shift
          ;;
      -h|--host|--hos|--ho)
          SERVER="http://$2"
          DEPOSIT_URL="${SERVER}/${DEPOSIT_PATH}"
          shift
          ;;
      -p|--path)
          DEPOSIT_PATH="$2"
          DEPOSIT_URL="${SERVER}/${DEPOSIT_PATH}"
          shift
          ;;
      -u|--url|--ur|--u)
          DEPOSIT_URL="$2"
          shift
          ;;
      -s|--separate-outputs|--separate)
          SEPARATE_OUT_DIR="$2"
          shift
          ;;
      --print-timestamps)
          TIMESTAMPS=true
          ;;
      --no-print-timestamps|--dont-print-timestamps)
          TIMESTAMPS=false
          ;;
      -d|--deposition-type)
          DEPOSITION_TYPE="$2"
          shift
          ;;
      -r|--read-stdin)
          READ_STDIN=true
          ;;
      -D|--debug|--debu|--deb|--de|--d)
          DEBUG=true
          ;;
      -D-|--no-debug|--no-debu|--no-deb|--no-de|--no-d)
          DEBUG=false
          ;;
      --output-mode)
          OUTPUT_MODE="$2"
          shift
          ;;
      -P|--progress|--progres|--progre|--progr|--prog|--pro|--pr|--p)
          PROGRESS=true
          ;;
      -P-|--no-progress|--no-progres|--no-progre|--no-progr|--no-prog|\
      --no-pro|--no-pr|--no-p)
          PROGRESS=false
          ;;
      --replace)
          REPLACE=true
          ;;
      --release-data|--release-dat|--release-da|--release-d|--release)
          RELEASE=true
          ;;
      --log-message|--log-messag|--log-messa|--log-mess|--log-mes|\
      --log-me|--log-m|--log|--lo|--l|--log-msg)
          LOG_MESSAGE="$2"
          shift
          ;;
      --ignore-unpaired-datablocks|--ignore-unpaired|\
      --silent)
          STRICT=false
          ;;
      --use-wipe|--use-wip|--use-wi|--use-w)
          USE_WIPE=true
          ;;
      --no-use-wipe|--no-use-wip|--no-use-wi|--no-use-w|--no-use-|--no-use|\
      --no-us|--no-u|\
      --use-rm|--use-r|\
      --dont-use-wipe|--dont-use-wip|--dont-use-wi|--dont-use-w)
          USE_WIPE=false
          ;;
      --no-ignore-unpaired-datablocks|--no-ignore-unpaired|\
      --dont-ignore-unpaired-datablocks|--dont-ignore-unpaired|\
      --strict)
          STRICT=true
          ;;
      --require-only-doi|--require-only-do|--require-only-d)
          REQUIRE_ONLY_DOI_ARG='-F doi_only=1'
          ;;
      --require-full-bibliography|--require-full-bibliograph|\
      --require-full-bibliograp|--require-full-bibliogra|\
      --require-full-bibliogr|--require-full-bibliog|\
      --require-full-biblio|--require-full-bibli|--require-full-bibl|\
      --require-full-bib|--require-full-bi|--require-full-b|\
      --require-full|--require-ful|--require-fu|--require-f)
          REQUIRE_ONLY_DOI_ARG=''
          ;;
      --common-cif-header|--common-cif-heade|--common-cif-head|\
      --common-cif-hea|--common-cif-he|--common-cif-h|--common-cif|\
      --common-ci|--common-c|--common|--commo|--comm|--com)
          CIF_HEADER=common
          ;;
      --iucr-cif-header|--iucr-cif-heade|--iucr-cif-head|--iucr-cif-hea|\
      --iucr-cif-h|--iucr-cif|--iucr-ci|--iucr-c|--iucr|--iuc|--iu|--i)
          CIF_HEADER=iucr
          ;;
      --crystaleye-cif-header|--crystaleye-cif-heade|--crystaleye-cif-head|\
      --crystaleye-cif-hea|--crystaleye-cif-he|--crystaleye-cif-h|\
      --crystaleye-cif|--crystaleye-ci|--crystaleye-c|\
      --crystaleye|--crystaley|--crystale|--crystal|--crysta|--cryst|--crys|\
      --cry|--cr)
          CIF_HEADER=crystaleye
          ;;
      --amcsd-cif-header|--amcsd-cif-heade|--amcsd-cif-head|--amcsd-cif-hea|\
      --amcsd-cif-h|--amcsd-cif|--amcsd-ci|--amcsd-c|\
      --amcsd|--amcs|--amc|--am|--a)
          CIF_HEADER=amcsd
          ;;
      --script|--scrip|--scri|--scr|--sc|--s)
          curl="script curl"
          ;;
      --run|--ru|--r)
          curl=curl
          ;;
      --options|--option|--optio|--opti|--opt|--op|--o)
          echo "${BASENAME}:: The '--options' option is a placeholder."
          echo "${BASENAME}:: It should be replaced by one of the following options:"
          awk '/#\* OPTIONS:/,/#\*\*/ {
                  sub("OPTIONS:", "");
                  sub("^ *#[*]?[*]?", "");
                  gsub("\\$0","'"$0"'");
                  print $0
              }' "$0"
          exit
          ;;
      --help|--hel|--he|--usage)
          gawk '/^#\*/,/^#\*\*/ {
                  sub("^ *#[*]?[*]?", "");
                  gsub("\\$0","'"$0"'");
                  print $0
              }' "$0"
          exit
          ;;
      --version)
          "$(dirname "$0")/cod-tools-version"
          exit
          ;;
      -*) echo "${BASENAME}:: ERROR, unknown option '$1'." >&2 ; exit 1 ;;
      *)  CIF_FILES="${CIF_FILES} $1" ;;
    esac
    shift
done

## echo ${CIF_FILES}
eval set -- "${CIF_FILES}"

if [ "$(echo "${CIF_FILES}" | wc -w)" -gt 1 ] && [ -n "${HKL_FILES}" ]
then
    echo "${BASENAME}:: ERROR, cannot accept Fobs files together with \
more than one CIF file." >&2
    exit 1
fi

if [ ${DEBUG} = true ]
then
    DEBUG_ARG='-F debug=1'
else
    DEBUG_ARG=''
fi

if [ ${PROGRESS} = true ]
then
    PROGRESS_ARG='-F progress=1'
else
    PROGRESS_ARG=''
fi

if [ ${REPLACE} = true ]
then
    REPLACE_ARG='-F replace=1'
else
    REPLACE_ARG=''
fi

if [ ${RELEASE} = true ]
then
    RELEASE_ARG='-F release=1'
else
    RELEASE_ARG=''
fi

if [ ${STRICT} = true ]
then
    STRICT_ARG='-F strict=1'
else
    STRICT_ARG='-F silent=1'
fi

if [ -n "${CIF_HEADER}" ]
then
    CIF_HEADER_ARG="-F header=${CIF_HEADER}"
else
    CIF_HEADER_ARG=''
fi

test -z "${TMP_DIR}" && TMP_DIR="/tmp"
TMP_DIR="${TMP_DIR}/tmp-${BASENAME}-$$"
umask 077
mkdir "${TMP_DIR}"

TMP_USERNAME="${TMP_DIR}/username"
TMP_PASSWRD="${TMP_DIR}/password"
TMP_EMAIL="${TMP_DIR}/user_email"
TMP_AUTHOR_NAME="${TMP_DIR}/author_name"
TMP_AUTHOR_EMAIL="${TMP_DIR}/author_email"
TMP_JOURNAL="${TMP_DIR}/journal"
TMP_CLIENT_IP="${TMP_DIR}/web_client_ip"
TMP_BYPASS_PASSWD="${TMP_DIR}/bypass_password"
TMP_MESSAGE="${TMP_DIR}/message"
TMP_DEPOSIT_LIST="${TMP_DIR}/deposition_list"

perl -lne "print \$1 if /^username=(.*)/" < "${INPUT_CONFIG}" > "${TMP_USERNAME}"
perl -lne "print \$1 if /^password=(.*)/" < "${INPUT_CONFIG}" > "${TMP_PASSWRD}"
perl -lne "print \$1 if /^journal=(.*)/"  < "${INPUT_CONFIG}" > "${TMP_JOURNAL}"
perl -lne "print \$1 if /^user_email=(.*)/"   < "${INPUT_CONFIG}" > "${TMP_EMAIL}"
perl -lne "print \$1 if /^author_name=(.*)/"  < "${INPUT_CONFIG}" > "${TMP_AUTHOR_NAME}"
perl -lne "print \$1 if /^author_email=(.*)/" < "${INPUT_CONFIG}" > "${TMP_AUTHOR_EMAIL}"
perl -lne "print \$1 if /^web_client_ip=(.*)/"   < "${INPUT_CONFIG}" > "${TMP_CLIENT_IP}"
perl -lne "print \$1 if /^bypass_check_password=(.*)/" < "${INPUT_CONFIG}" > "${TMP_BYPASS_PASSWD}"
perl -lne "print \$1 if /^message=(.*)/" < "${INPUT_CONFIG}" > "${TMP_MESSAGE}"
HOLD_PERIOD=$(perl -lne "print \$1 if /^hold_period=(.*)/" < "${INPUT_CONFIG}")

if [ -n "${LOG_MESSAGE}" ]
then
    echo "${LOG_MESSAGE}" > "${TMP_MESSAGE}"
fi

if [ -s "${TMP_JOURNAL}" ]
then
    JOURNAL_ARG="-F journal=<${TMP_JOURNAL}"
else
    JOURNAL_ARG=''
fi

if [ -s "${TMP_EMAIL}" ]
then
    EMAIL_ARG="-F user_email=<${TMP_EMAIL}"
else
    EMAIL_ARG=''
fi

if [ -s "${TMP_AUTHOR_NAME}" ]
then
    AUTHOR_NAME_ARG="-F author_name=<${TMP_AUTHOR_NAME}"
else
    AUTHOR_NAME_ARG=''
fi

if [ -s "${TMP_AUTHOR_EMAIL}" ]
then
    AUTHOR_EMAIL_ARG="-F author_email=<${TMP_AUTHOR_EMAIL}"
else
    AUTHOR_EMAIL_ARG=''
fi

if [ -s "${TMP_CLIENT_IP}" ]
then
    WEB_CLIENT_IP_ARG="-F web_client_ip=<${TMP_CLIENT_IP}"
else
    WEB_CLIENT_IP_ARG=''
fi

if [ -s "${TMP_BYPASS_PASSWD}" ]
then
    BYPASS_PASSWD_ARG="-F bypass_check_password=<${TMP_BYPASS_PASSWD}"
else
    BYPASS_PASSWD_ARG=''
fi

if [ -s "${TMP_MESSAGE}" ]
then
    MESSAGE_ARG="-F message=<${TMP_MESSAGE}"
else
    MESSAGE_ARG=''
fi

if [ -n "${HOLD_PERIOD}" ]
then
    HOLD_PERIOD_ARG="-F hold_period=${HOLD_PERIOD}"
else
    HOLD_PERIOD_ARG=''
fi

DEPOSIT_LOGGER="cat"

if [ ${READ_STDIN} = true ]
then
    cat "$@" > "${TMP_DEPOSIT_LIST}"
else
    if [ -n "${HKL_FILES}" ]
    then
        echo "${CIF_FILES}" "${HKL_FILES}" > "${TMP_DEPOSIT_LIST}"
    else
        perl -e 'while( @ARGV ) { print shift( @ARGV ), "\n"; }' \
            "${CIF_FILES}" > "${TMP_DEPOSIT_LIST}"
    fi
fi

cat "${TMP_DEPOSIT_LIST}" | \
while read -r line
do
    CIF=$(echo "$line" | cut -d "${SEPARATOR}" -f 1)
    HKL=$(echo "$line" | cut -d "${SEPARATOR}" -f 2-)
    if [ ${TIMESTAMPS} = true ]
    then
        echo "${BASENAME}: ${CIF}: NOTE, deposition started at $(date)."
    fi
    CIF_ARG="-F cif=@${CIF};filename=$(basename "${CIF}")"
    HKL_ARG=""
    if [ "${line}" != "${line/${SEPARATOR}/}" ]
        then HKL_ARG=$(echo "${HKL}" \
                        | perl -lne 'use File::Basename;
                                     for my $hkl (split( "'"${SEPARATOR}"'", $_ ) ) {
                                         print "-F hkl=\@${hkl};filename=",
                                               basename( $hkl );
                                     }')
    fi
    if [ -n "${SEPARATE_OUT_DIR}" ]
    then
        DEPOSIT_LOGGER="cat > \"${SEPARATE_OUT_DIR}/$(basename "${CIF}" .cif).html\""
    fi
    # Variables like ${STRICT_ARG}, ${CIF_ARG} and ${HKL_ARG} must not be
    # double quoted since their values also contain the '-F' curl option. 
    # shellcheck disable=SC2086
    ${curl} \
        --silent \
        --show-error \
        --user-agent "${BASENAME}" \
        ${DEBUG_ARG} \
        ${PROGRESS_ARG} \
        ${REPLACE_ARG} \
        ${MESSAGE_ARG} \
        ${RELEASE_ARG} \
        ${STRICT_ARG} \
        ${REQUIRE_ONLY_DOI_ARG} \
        ${CIF_HEADER_ARG} \
        -F "output_mode=${OUTPUT_MODE}" \
        -F "username=<${TMP_USERNAME}" \
        -F "password=<${TMP_PASSWRD}" \
        ${JOURNAL_ARG} \
        ${EMAIL_ARG} \
        ${AUTHOR_NAME_ARG} \
        ${AUTHOR_EMAIL_ARG} \
        ${WEB_CLIENT_IP_ARG} \
        ${BYPASS_PASSWD_ARG} \
        ${HOLD_PERIOD_ARG} \
        ${CIF_ARG} \
        ${HKL_ARG} \
        -F "deposition_type=${DEPOSITION_TYPE}" \
        "${DEPOSIT_URL}" \
        2>&1 \
        | eval "${DEPOSIT_LOGGER}"
    if [ ${TIMESTAMPS} = true ]
    then
        echo "${BASENAME}: ${CIF}: NOTE, deposition finished at $(date)."
    fi
done

if which wipe > /dev/null 2>&1 && \
   [ ${USE_WIPE} = true ]
then
    wipe -rf "${TMP_DIR}"
else
    rm -rf "${TMP_DIR}"
fi