source: TOOLS/PACK_IPSL/archive_restart.sh @ 3632

#!/bin/bash


#on veut archiver les fichiers restart
#Dans le cadre du demenagement ils seront renommes
#ATM/Restart/****.nc
#OCE/Restart/****.nc
#...
#sous la forme
#RESTART/ATM_****.nc
#RESTART/OCE_****.nc
#Attention de bien conserver les dates d'origines 
#
#Dans le cadre de la production nouvelle chaine
#ils auront directement le bon nom
#
#On veut egalement que l'archive soit par periode de temps 
# cette periode est definie par la taille du pack donc connue 
# en entree du script (1M, 1Y, 5Y, 10Y etc...) 
#


# $1= PATH/config_card d'une simulation avec 
# TagName/JobName/ExperimentName/SpaceName/DateBegin/DateEnd/PATH_SIMUL
# $2=period_pack en M/m ou Y/y

config_card=${1}
LISTE_DMNFS=${2}


export JOB_DIR=${JOB_DIR:-${TMP_MIGR_DATA}}
export IGCM_DEM=${IGCM_DEM:-${SCRATCHDIR}/IGCM_DEM}
export IGCM_TMP=${IGCM_TMP:-${IGCM_DEM}/tmp}
mkdir -p ${IGCM_DEM}
mkdir -p ${IGCM_TMP}
clean_tmp=${clean_tmp:-yes}
EXE_DIR=$(dirname ${0})
SCRIPT_NAME=$(basename ${0} )

source ${EXE_DIR}/DEM_utilities.sh

# gpdebug : verif si le present script a ete execute jusqu'au bout #########
MonitoringInfo="${SCRIPT_NAME}-->OK"
simuMonitoring_check "$MonitoringInfo" "$config_card"

DEM_log -0 "Demarrage de ${SCRIPT_NAME}"


#### 0- lecture de config.card 
#### on veut les renseignements suivants : 
#### JobName, PATH_SIMUL, DateBegin et DateEnd
eval $(grep JobName ${config_card}) 
eval $(grep PATH_SIMUL_FULL ${config_card}) 
eval $(grep DateBegin ${config_card}) 
eval $(grep DateEnd ${config_card}) 
eval $(grep IGCM_DEM_SIMU ${config_card})


#### Preparation de l'environnement
# repertoire dans lequel seront renommes les fichiers de restart
STORE_RESTART=$IGCM_DEM_SIMU/RESTART
mkdir -p ${STORE_RESTART}

# repertoire dans lequel nous stockerons les listes de restart a archiver
RESTART_TAR=$IGCM_DEM_SIMU/restart_tar
mkdir -p ${RESTART_TAR}

# lecture de periodpack dans ${IGCM_DEM_SIMU}/period_pack.txt
period_pack=`awk '{print $0} ' ${IGCM_DEM_SIMU}/period_pack.txt`

DEM_log -1 "Period pack = ${period_pack}" 
#### 1- renommer les fichiers ######
DEM_log -1 "Renommer les fichiers de Restart"

cd ${STORE_RESTART}
if [ "x${STORE_RESTART}" != "x" ]
then
    rm -rf *
fi
# On les copie sur le scratchdir en changeant le nom et conservant la date
awk '{print $2}' ${IGCM_DEM_SIMU}/liste_restart_files_config.txt > ${IGCM_TMP}/${$}.liste_restart_files.txt
#grep ${PATH_SIMUL_FULL} ${LISTE_DMNFS} | grep Restart | awk '{print $3}' | sed '/^.*Restart$/d' >> ${IGCM_TMP}/${$}.liste_restart_files.txt

  
gawk -F/ '{dest_name=$(NF-2)("_") $(NF); system("ln -s " $0 " "dest_name)}' ${IGCM_TMP}/${$}.liste_restart_files.txt

ls -l ${STORE_RESTART} > ${IGCM_TMP}/${$}.new_liste_restart_files.txt

### creation de liste en fonction du type de restart
   if [ "x${ON_IDRIS}" == "xno" ]
   then
       awk '{if (NR >1) print $9}' ${IGCM_TMP}/${$}.new_liste_restart_files.txt > ${IGCM_TMP}/${$}.new_liste_restart_files.txt_tmp
   else
       awk '{if (NR >1) print $8}' ${IGCM_TMP}/${$}.new_liste_restart_files.txt > ${IGCM_TMP}/${$}.new_liste_restart_files.txt_tmp       
   fi
sort ${IGCM_TMP}/${$}.new_liste_restart_files.txt_tmp > ${IGCM_TMP}/${$}.order_new_liste_restart_files.txt
ref=`awk -F_ '{if (NR==1) print $1}' ${IGCM_TMP}/${$}.order_new_liste_restart_files.txt`

i=1
for new_file in `awk  '{print $0}' ${IGCM_TMP}/${$}.order_new_liste_restart_files.txt`; do 
    new_ref=$( echo ${new_file} | sed -e "s/_.*//") 

    if [ ${new_ref} == ${ref} ]; then
        echo ${new_file} >> ${IGCM_TMP}/${$}.restart_list_${i}.txt
    else
        ref=${new_ref}
        (( i = $i + 1 )) 
        echo ${new_file} >> ${IGCM_TMP}/${$}.restart_list_${i}.txt
    fi
done



j=1
while [[ $j -le $i  ]]; do 

### 2- creer les listes de fichiers par periode ####
    DEM_log -1 "Creation des listes de fichiers par periode"
### 2.1- extraire l'annee et le mois de DateBegin ###
    date_begin=${DateBegin}
    date_end=${DateEnd}

    year_begin=$(echo ${date_begin} | cut -c 1-4 )
    month_begin=$(echo ${date_begin} | cut -c 5-6 )
    year_end=$(echo ${date_end} | cut -c 1-4 )
    month_end=$(echo ${date_end} | cut -c 5-6 )


### 2.2- creation des listes ###
    date_begin_pack=${date_begin}
    date_end_simulation=${date_end}

    month_begin_pack=${month_begin}
    year_begin_pack=${year_begin}

    number_pack=1
    echo date_end_simulation : $date_end_simulation
    while [ ${date_begin_pack} -le ${date_end_simulation} ]
    do 
        
        echo "On calcule la date de fin du pack "

        month_end_pack=${month_begin_pack}
        year_end_pack=`expr ${year_begin_pack} + ${period_pack}`


        date_end_pack=${year_end_pack}${month_end_pack}01   
        echo date_end_pack $date_end_pack
        
        for file in `awk '{print $1}' ${IGCM_TMP}/${$}.restart_list_${j}.txt`; do   
            extract_date_file=`echo ${file}  | sed -e "s/.*\${JobName}_//"`
            date_file=`echo ${extract_date_file}| sed 's/\([0-9]\{8\}\)_.*$/\1/g'`
            
            if [  ${date_file} -lt ${date_end_pack} ]
            then
                if [ ${date_file} -ge ${date_begin_pack} ]
                then 
                    echo ${file} >> ${IGCM_TMP}/${$}.liste_pack_restart_${number_pack}.txt
                    grep -v ${file} ${IGCM_TMP}/${$}.restart_list_${j}.txt > ${IGCM_TMP}/${$}.restart_list_${j}.txt.tmp
                    mv ${IGCM_TMP}/${$}.restart_list_${j}.txt.tmp ${IGCM_TMP}/${$}.restart_list_${j}.txt
                fi
                
            fi
            if [  ${date_file} -gt ${date_end_pack} ]
            then 
                break
            fi
        done



        DEM_log -1 "On definit le datebegin de la  prochaine periode pack"
#on definit la periode pack suivante
        month_begin_pack=${month_end_pack}
        year_begin_pack=${year_end_pack}
        
        date_begin_pack=${year_begin_pack}${month_begin_pack}01
        
        number_pack=`expr ${number_pack} + 1`

    done

    ((j = $j + 1 ))
done

compt_pack=1
while [[ $compt_pack -le $number_pack ]]; do 
#on archive les restart - le nom de l'archive contiendra ${yearbegin}${monthbegin}_${yearend}${monthend} 
# end etant la date du dernier fichier contenu et non pas la premiere date suivant 
    DEM_log -1 "Recherche de la date de fin reel pour nommer l'archive "
#On classe par ordre alphabetique les fichiers contenus dans liste_pack_restart 
#la date du dernier fichier donnera la date de fin du pack
    if [[ -f ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt ]] ; then
        sort ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt > ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt.tmp

        first_file=$(head -n 1 ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt.tmp )
        extract_date_file=$( echo ${first_file}  | sed -e "s/.*${JobName}_//" )
        DateBegin_Restart=$(echo ${extract_date_file}| sed 's%\([0-9]\{8\}\)_.*$%\1%g' )
            
        last_file=$(tail -n 1 ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt.tmp )
        extract_date_file=$( echo ${last_file}  | sed -e "s/.*${JobName}_//" )
        DateEnd_Restart=$(   echo ${extract_date_file}| sed 's/\([0-9]\{8\}\)_.*$/\1/g' )
        mv ${IGCM_TMP}/${$}.liste_pack_restart_${compt_pack}.txt.tmp ${RESTART_TAR}/${JobName}_${DateBegin_Restart}_${DateEnd_Restart}_restart.list
    fi

compt_pack=`expr ${compt_pack} + 1`
done


cd ${EXE_DIR}

# gpdebug : le script n'a pas rencontre d'erreur. On l'inscrit dans le fichier de suivi
simuMonitoring_OK "$MonitoringInfo" "$config_card"