#!/bin/bash

# source global variable
source 0_export.bash
source activate ${PY_ENV_NAME}
export GTIFF_SRS_SOURCE='EPSG'

YEAR_START=${1:-${MODEL_YEAR_START}}
YEAR_STOP=${2:-${MODEL_YEAR_STOP}}
PREPROCESS_TIME_INDEPENDENT_SWITCH=${3:-${PREPROCESS_TIME_INDEPENDENT_SWITCH}}
PREPROCESS_MODEL_CMIP6_SWITCH=${4:-${PREPROCESS_MODEL_CMIP6_SWITCH}}
PREPROCESS_MODEL_ISIMIP3B_SWITCH=${5:-${PREPROCESS_MODEL_ISIMIP3B_SWITCH}}

if [ "${PREPROCESS_TIME_INDEPENDENT_SWITCH}" == True ];then
    echo "  %Status: preparing time independent inputs"
    [ ! -d ${TIME_INDEPENDENT_REGION_INPUT_DIR} ] && mkdir -p ${TIME_INDEPENDENT_REGION_INPUT_DIR}
    python ${PY_SCRIPT_PREPARE_TIME_INDEPENDENT_INPUT} -i ${TIME_INDEPENDENT_INPUT_DIR} -o ${TIME_INDEPENDENT_REGION_INPUT_DIR} -au ${TIME_INDEPENDENT_AUX_DIR} -s ${TIME_INDEPENDENT_SRAD_DIR} -xmn ${XMIN} -xmx ${XMAX} -ymn ${YMIN} -ymx ${YMAX} -bf ${XYBUFFER} -pg
fi

if [ "${PREPROCESS_MODEL_CMIP6_SWITCH}" == True ];then
    [ ! -d ${MODEL_CMIP6_REGION_DIR} ] && mkdir -p ${MODEL_CMIP6_REGION_DIR}
    echo "  %Status: preparing CMIP6 data of ${MODEL_NAME}"
    ZG_TA_FILE_LIST=$(find -L ${MODEL_CMIP6_INPUT_DIR} -name "*.nc" | grep 'ta_\|zg_')
    if  grep -q '_day_' <<< ${ZG_TA_FILE_LIST} && grep -q '_Eday_' <<< ${ZG_TA_FILE_LIST};then
        echo "  %Error: Found variables ta and zg with table_id day AND Eday."
        echo "  %Error: You should provide only one kind of table_id for these variables,"
        echo "  %Error: to avoid inconsistencies in pressure levels."
        exit 1
    fi
    for VARIABLE in ${MODEL_CMIP6_VARIABLE_LIST};do
        echo "  %Status:     working on variable ${VARIABLE}"
        FILE_LIST=$(find -L ${MODEL_CMIP6_INPUT_DIR} -name "*${VARIABLE}_*.nc*" | grep 'day_' | grep "${MODEL_EXPERIMENT}")
        if [  "${FILE_LIST}" == "" ];then
            echo "  %Warning:    no file found for ${VARIABLE}"
        else
            # iterate over years
            # each output file contains one variable and one year
            for YEAR in $(seq ${YEAR_START} ${YEAR_STOP});do
                echo "  %Status:         year = ${YEAR}"

                # make the domain to cut out larger because the cmip grid is
                # coarser and we don't want to miss any grid cells
                XMIN_LARGE=$(python -c "print('{0}'.format(${XMIN} - 3.0))")
                XMAX_LARGE=$(python -c "print('{0}'.format(${XMAX} + 3.0))")
                YMIN_LARGE=$(python -c "print('{0}'.format(${YMIN} - 3.0))")
                YMAX_LARGE=$(python -c "print('{0}'.format(${YMAX} + 3.0))")

                # filter FILE_LIST for files containing only this year
                YEAR_FILE_LIST=$(filter_files_by_year_range "${FILE_LIST}" ${YEAR} ${YEAR})

                # define output filename and auxiliary filename
                FILE_OUT=${MODEL_CMIP6_REGION_DIR}/${VARIABLE}_day_${MODEL_NAME}_${MODEL_EXPERIMENT}_${MODEL_REALIZATION}_${YEAR}.nc
                AUX_FILE=$(dirname ${FILE_OUT})/aux_$(random_number).nc

                # cut domain
                cdo -L -s -f nc4 -k grid -z zip -sellonlatbox,${XMIN_LARGE},${XMAX_LARGE},${YMIN_LARGE},${YMAX_LARGE} -selyear,${YEAR} -mergetime ${YEAR_FILE_LIST} ${AUX_FILE}
                # change calendar to proleptic_gregorian
                python -Bu ${PY_SCRIPT_CHANGE_CALENDAR} ${AUX_FILE} ${FILE_OUT} proleptic_gregorian
                rm -f ${AUX_FILE}
            done
        fi
    done
fi

if [ "${PREPROCESS_MODEL_ISIMIP3B_SWITCH}" == True ];then
    [ ! -d ${MODEL_ISIMIP3B_REGION_DIR} ] && mkdir -p ${MODEL_ISIMIP3B_REGION_DIR}
    echo "  %Status: preparing ISIMIP3b data of ${MODEL_NAME}"
    for VARIABLE in ${MODEL_ISIMIP3B_VARIABLE_LIST};do
        echo "  %Status:     working on variable ${VARIABLE}"
        FILE_LIST=$(find -L ${MODEL_ISIMIP3B_INPUT_DIR} -name "*${VARIABLE}_*.nc*" | grep '_daily_' | grep "${MODEL_EXPERIMENT}")
        if [  "${FILE_LIST}" == "" ];then
            echo "  %Warning:    no file found for ${VARIABLE}"
        else
            # iterate over years
            # each output file contains one variable and one year
            for YEAR in $(seq ${YEAR_START} ${YEAR_STOP});do
                echo "  %Status:         year = ${YEAR}"

                # filter FILE_LIST for files containing only this year
                YEAR_FILE_LIST=$(filter_files_by_year_range "${FILE_LIST}" ${YEAR} ${YEAR})
                # convert MODEL_NAME to lower case (to be inline with the ISIMIP3b-naming)
                MODEL_NAME_LOWER=$(tr '[:upper:]' '[:lower:]' <<< ${MODEL_NAME})

                # define output filename and auxiliary filename
                FILE_OUT=${MODEL_ISIMIP3B_REGION_DIR}/${VARIABLE}_day_${MODEL_NAME_LOWER}_${MODEL_EXPERIMENT}_${MODEL_REALIZATION}_${YEAR}.nc
                AUX_FILE=$(dirname ${FILE_OUT})/aux_$(random_number).nc

                # cut domain
                cdo -L -s -f nc4 -k grid -z zip -sellonlatbox,${XMIN},${XMAX},${YMIN},${YMAX} -selyear,${YEAR} -mergetime ${YEAR_FILE_LIST} ${AUX_FILE}
                # change calendar to proleptic_gregorian
                python -Bu ${PY_SCRIPT_CHANGE_CALENDAR} ${AUX_FILE} ${FILE_OUT} proleptic_gregorian
                rm -f ${AUX_FILE}
            done
        fi
    done
fi