#!/bin/bash

# Script permettant de creer les listes de fichiers pour ncrcat
# si un trou est detecte dans la simulation alors cette liste sera 
# pour un tar

# $1= PATH/config_card d'une simulation avec 
# TagName/JobName/ExperimentName/SpaceName/DateBegin/DateEnd/PATH_SIMUL


# Pour commencer on travaille avec le fichier information_dmnfs
# /cscratch/cont003/p86mart/FROM_CCRT_INFORMATIONS_DMNFS/information_dmnfs_2011-09-21

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

# Declaration des variables

# ----
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"

# gpdebug : Execute la function errorReceive() si reception d'un signal TERM
trap 'DEM_errorReceive' TERM ########

# gpdebug : on augmente la liste des proc à détruire en cas d'erreur bloquante du présent proc
export listPID="$$ ${listPID}" #########

DEM_log -0 "Demmarrage de ${SCRIPT_NAME}"
cd ${JOB_DIR}
# ----

DEM_log -0 "config_card : ${config_card}"
DEM_log -0 "LISTE_DMNFS : ${LISTE_DMNFS}"

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})

SIMUL=${PATH_SIMUL_FULL}/${JobName}


DEM_state=$( DEM_read_state ${IGCM_DEM}/config_card.liste ${config_card} )
# gpdebug : si c'est une premiere exec (pas une reprise)
if [[ "x${execEveryStep}" == "1" ]] && [[ "${DEM_state}" != "FindSizePackDone" ]] ; then
    DEM_errorSend "${SCRIPT_NAME}:${LINENO}:Etat incorrect dans ${IGCM_DEM}/config_card.liste"
fi


DEM_write_state ${IGCM_DEM}/config_card.liste ${config_card} WriteListPackRunning

# creation des repertoires with_tar et with_ncrcat
OUTPUT_NCRCAT=${IGCM_DEM_SIMU}/output_ncrcat
OUTPUT_TAR=${IGCM_DEM_SIMU}/output_tar

mkdir -p ${OUTPUT_NCRCAT}
mkdir -p ${OUTPUT_TAR}

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

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

awk '{print $2}' ${IGCM_DEM_SIMU}/liste_output_files_config.txt >  ${IGCM_DEM_SIMU}/info_Output_light.txt

# on cree la liste des fichiers pour chaque type_file
for type_file in $( awk '{print $0}' ${IGCM_DEM_SIMU}/info_name_file.txt ) ; do 
   #sed -e "/${type_file}/!d" ${IGCM_DEM_SIMU}/info_Output_light.txt > ${IGCM_TMP}/${$}.info_${type_file}.txt
    grep ${type_file} ${IGCM_DEM_SIMU}/info_Output_light.txt > ${IGCM_TMP}/${$}.info_${type_file}.txt
    sort -u ${IGCM_TMP}/${$}.info_${type_file}.txt > ${IGCM_TMP}/${$}.tmp_info_${type_file}.txt.tmp
    cp ${IGCM_TMP}/${$}.tmp_info_${type_file}.txt.tmp ${IGCM_TMP}/${$}.info_${type_file}.txt
    [[ "${clean_tmp}" = "yes" ]] && rm -f ${IGCM_TMP}/${$}.tmp_info_${type_file}.txt.tmp
done

# 3- Maintenant pour chaque type de fichier de sortie nous allons creer la liste
# des fichiers contenus dans chaque pack

# Recherche des dates du premier pack

date_begin_pack=${DateBegin}
date_end_simulation=${DateEnd}

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

number_pack=1

#
date_begin_pack=`echo $date_begin_pack|sed 's/^0*//'`
date_end_simulation=`echo $date_end_simulation|sed 's/^0*//'`

while [[ ${date_begin_pack} -le ${date_end_simulation}  ]]
do 

   month_end_pack=${month_begin_pack}
   year_end_pack=$(( ${year_begin_pack} +${period_pack} ))
   
   date_end_pack=${year_end_pack}${month_end_pack}01   

   DEM_log -2 "Entree boucle 3 : year_begin_pack month_begin_pack year_end_pack month_end_pack : ${year_begin_pack} ${month_begin_pack} ${year_end_pack} ${month_end_pack}"
   DEM_log -2 "Entree boucle 3 : date_end_pack : ${date_end_pack}"
   
   for type_file in `awk '{print $0}' ${IGCM_DEM_SIMU}/info_name_file.txt`
   do 
       DEM_log -2 "Boucle 3 : ${type_file} ${number_pack}"
       taille=0

       # On passe en revue chaque type de fichier 
       for file in `awk '{print $0}' ${IGCM_TMP}/${$}.info_${type_file}.txt`
       do 
           # Extraire la date de debut et la date de fin de la periode d'output traitee par le fichier    
           extract_date_file=$( echo ${file}             | sed -e "s%.*${JobName}_%%" )
           date_debut=$(        echo ${extract_date_file}| sed 's%\([0-9]\{8\}\)_.*$%\1%g' )
           date_fin=$(          echo ${extract_date_file}| sed 's%.*_\([0-9]\{8\}\)_.*$%\1%g' )
	   
           if [  ${date_debut} -lt ${date_end_pack} ]
           then 
               if [ ${date_fin} -ge ${date_begin_pack} ]
               then 
		   echo ${file} >> ${OUTPUT_NCRCAT}/ncrcat_${type_file}_${date_begin_pack}_${date_end_pack}.list
		   echo ${date_debut} | cut -c 1-6 >>  ${IGCM_TMP}/${$}.date_${type_file}.txt
		   grep -v ${file} ${IGCM_TMP}/${$}.info_${type_file}.txt > ${IGCM_TMP}/${$}.info_${type_file}.txt.tmp
		   cp ${IGCM_TMP}/${$}.info_${type_file}.txt.tmp ${IGCM_TMP}/${$}.info_${type_file}.txt
		   [[ "${clean_tmp}" = "yes" ]] && rm -f ${IGCM_TMP}/${$}.info_${type_file}.txt.tmp
               fi
	   else 
	       break
           fi
	   
       done
       if [[ -f ${IGCM_TMP}/${$}.date_${type_file}.txt ]] ; then
	   sort -u ${IGCM_TMP}/${$}.date_${type_file}.txt > ${IGCM_TMP}/${$}.date_tmp.txt
	   mv ${IGCM_TMP}/${$}.date_tmp.txt ${IGCM_TMP}/${$}.date_${type_file}.txt
       else
	   DEM_log -2 "Pas de fichier cree :  ${IGCM_TMP}/${$}.date_${type_file}.txt - Date_end_pack : ${date_end_pack}"
       fi

#On renomme le fichier ncrcat avec la date reelle de fin et non pas la premiere date suivant la fin 
#on renomme egalement en fonction de la date reelle de debut pour le cas des annees < 1000 qui perdent leurs zeros en cour de route
       if [[ -f ${OUTPUT_NCRCAT}/ncrcat_${type_file}_${date_begin_pack}_${date_end_pack}.list ]]; then 
	   sort ${OUTPUT_NCRCAT}/ncrcat_${type_file}_${date_begin_pack}_${date_end_pack}.list > ${IGCM_TMP}/${$}.ncrcat.tmp

	   first_file=$(head -n 1 ${IGCM_TMP}/${$}.ncrcat.tmp )
	   extract_date_file=$( echo ${first_file}  | sed -e "s/.*${JobName}_//" )
	   date_debut=$(echo ${extract_date_file}| sed 's%\([0-9]\{8\}\)_.*$%\1%g' )

	   last_file=$(tail -n 1 ${IGCM_TMP}/${$}.ncrcat.tmp )
	   extract_date_file=$( echo ${last_file}  | sed -e "s/.*${JobName}_//" )
	   date_fin=$(echo ${extract_date_file}| sed 's%.*_\([0-9]\{8\}\)_.*$%\1%g' )
	   mv ${OUTPUT_NCRCAT}/ncrcat_${type_file}_${date_begin_pack}_${date_end_pack}.list ${OUTPUT_NCRCAT}/${JobName}_${date_debut}_${date_fin}_${type_file}.list
       fi
   done

   month_begin_pack=${month_end_pack}
   year_begin_pack=${year_end_pack}
   date_begin_pack=${year_begin_pack}${month_begin_pack}01
   
   number_pack=$(( ${number_pack} + 1 ))

   DEM_log -2 "Sortie boucle : year_begin_pack month_begin_pack year_end_pack month_end_pack : ${year_begin_pack} ${month_begin_pack} ${year_end_pack} ${month_end_pack}"
   DEM_log -2 "Sortie boucle : date_begin_pack : ${date_begin_pack}"
   
done


## Recherche des trous eventuels dans cette simulation

## Creation de la liste des annees comprisent entre DateBegin et DateEnd >>  ${IGCM_TMP}/${$}.liste_date.txt
force_sequence (){
    awk '{
        if (NR==1) {avant=$1; print $1}
        else { for (i=avant+1 ; i<=$1 ; i++) printf "%04i\n", i}}'
}

DEM_log -1 "year_begin - year_end : ${year_begin} - ${year_end}"

echo ${year_begin} > ${IGCM_TMP}/${$}.liste_date.txt
echo ${year_end}  >> ${IGCM_TMP}/${$}.liste_date.txt

force_sequence < ${IGCM_TMP}/${$}.liste_date.txt > ${IGCM_TMP}/${$}.tmp ; mv ${IGCM_TMP}/${$}.tmp ${IGCM_TMP}/${$}.liste_date.txt


## il faut savoir si les fichiers d'output sont annuels ou mensuels ? 
file=$( head -n 1 ${IGCM_DEM_SIMU}/liste_output_files_config.txt )

extract_date_file=$( echo ${file}             | sed -e "s%.*${JobName}_%%" )
date_debut=$(        echo ${extract_date_file}| sed 's%\([0-9]\{8\}\)_.*$%\1%g')
date_fin=$(          echo ${extract_date_file}| sed 's%.*_\([0-9]\{8\}\)_.*$%\1%g')
month_begin=$(       echo ${date_debut}       | cut -c 5-6 )
month_end=$(         echo ${date_fin}         | cut -c 5-6 )
year_debut=$(       echo ${date_debut}       | cut -c 1-4 )
year_fin=$(       echo ${date_fin}       | cut -c 1-4 )
DEM_log -1 "Recherche de la frequence"
if [ ${month_end} -eq ${month_begin} ] 
then 
    DEM_log -1 "Mensuel"
    while read an ; do
        for month in 1 2 3 4 5 6 7 8 9 10 11 12 ; do
            printf "%s%02d\n" ${an} ${month}  >> ${IGCM_TMP}/${$}.toutes_dates
	done
    done < ${IGCM_TMP}/${$}.liste_date.txt
    [[ "${clean_tmp}" = "yes" ]] && rm -f ${IGCM_TMP}/${$}.liste_date.txt
else
    freqwriteyear=0
#la frequence peut être de plusieurs mois (<1 an) 
#ou de plusieurs années 
#attention une simulation ne commence pas obligatoirement au 1er janvier
    if [[ ${year_debut} -eq ${year_fin} ]]; then 
	freqwritemonth=$(( ${month_end} - ${month_begin} + 1 ))
    else
	freqwritemonth=$(( 12 + ${month_end} - ${month_begin} + 1 ))
    fi
    if [[ $freqwritemonth -eq 12 ]]; then
	## frequence annuelle
	if [[ ${year_debut} -eq ${year_fin} ]]; then 
	    freqwriteyear=1
	else 
	    if [[ ${month_end} -gt {month_begin}  ]]; then 
		freqwriteyear=$(( ${year_fin} - ${year_debut} + 1 ))
	    else
		freqwriteyear=$(( ${year_fin} - ${year_debut} ))
	    fi
	fi
    fi 

    if [[ $freqwriteyear -eq 0 ]]; then 
	DEM_log -1 "mensuel avec sorties tous les  $freqwritemonth mois "
	while read an ; do
	    i=${month_begin}
	    while [[ ${i} -le 12 ]] ; do 
		printf "%s%02d\n" ${an} ${i}  >> ${IGCM_TMP}/${$}.toutes_dates
		i=$(( $i + $freqwritemonth ))
	    done
	done < ${IGCM_TMP}/${$}.liste_date.txt    
    else
	DEM_log -1 "annuel avec sorties tous les  $freqwriteyear month "
	i=${month_begin}
	y=${year_debut}
	while [[ $y -le ${year_end} ]]; do    # attention year_fin = annee de fin du premier fichier / year_end = annee de fin de la simulation 
		printf "%s%02d\n" ${y} ${i}  >> ${IGCM_TMP}/${$}.toutes_dates
		y=$(( $y + $freqwriteyear ))
	done
    fi 
	
fi 

# Maintenant il faut comparer la liste des dates avec la liste des dates des fichiers 
# et pour des dates manquantes transformer le fichier ncrcat_date correspondant en fichier 
# tar_date
DEM_log -1 "Recherche des dates manquantes"
for type_file in `awk '{print $0}' ${IGCM_DEM_SIMU}/info_name_file.txt`
do 
    DEM_log -2 "${type_file}"
    rm -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list
    
    if [ -f ${IGCM_TMP}/${$}.date_${type_file}.txt ] ; then
	join -v 1  ${IGCM_TMP}/${$}.toutes_dates ${IGCM_TMP}/${$}.date_${type_file}.txt > ${IGCM_DEM_SIMU}/${type_file}_manquant.list
	
        # On change le fichier ncrcat pour chaque date manquante en fichier tar
	if [[ -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list && -s ${IGCM_DEM_SIMU}/${type_file}_manquant.list ]]; then 
	    for date in `awk '{print $0}' ${IGCM_DEM_SIMU}/${type_file}_manquant.list` ; do
### AC : test si le fichier de list existe dans output_ncrcat (si non ca veut dire qu'il deja ete move dans output_tar) 
		nb_files=$(ls ${OUTPUT_NCRCAT} | grep ${JobName} | grep "${type_file}.list" | wc -l )
		if [ "x$nb_files" != "x0" ] ; then
		    ls -l ${OUTPUT_NCRCAT}/${JobName}_*_${type_file}.list > ${IGCM_TMP}/${$}.liste_ncrcat
		    for pack in $( awk '{print $9}' ${IGCM_TMP}/${$}.liste_ncrcat ) ; do 
			extract_date_file=$( echo ${pack}       | sed -e "s%.*${JobName}_%%" )
			date_debut=$( echo ${extract_date_file} | sed 's%\([0-9]\{8\}\)_.*$%\1%g'     | cut -c 1-6  )
			date_fin=$(   echo ${extract_date_file} | sed 's%.*_\([0-9]\{8\}\)_.*$%\1%g' | cut -c 1-6  )
			#on retire les eventuels zero pour eviter de tomber sur des dates de type 08 ou 09 
			date=`echo $date|sed 's/^0*//'`
			date_debut=`echo $date_debut|sed 's/^0*//'`
			date_fin=`echo $date_fin|sed 's/^0*//'`
			if [[ ${date} -ge ${date_debut} && ${date} -lt ${date_fin} ]]; then
			    mv ${pack} ${OUTPUT_TAR}/.
			    break 
			else
			    rm -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list
			fi
		    done
		fi
		[[ "${clean_tmp}" = "yes" ]] && rm -f ${IGCM_TMP}/${$}.liste_ncrcat
	    done
	elif [[ -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list && ! -s ${IGCM_DEM_SIMU}/${type_file}_manquant.list ]]; then
	    rm -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list
	fi
    fi
    rm -f ${IGCM_DEM_SIMU}/${type_file}_manquant.list
done

DEM_write_state ${IGCM_DEM}/config_card.liste ${config_card} WriteListPackDone

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

DEM_log -0 "Fin de ${SCRIPT_NAME}"

exit