#!/bin/bash

# this script has been written to run on La Palma on the machine data
#   i.e. paths are only working on this machine
# the script starts from the zipped files
#   this causes a delay until files are in the database
#   because they have to be rsynced and zipped first (see RsyncRawData.sh, ZipRawData.sh)

# missing
# entry in status table

# options: 
skipmd5sum="no" # fill md5 sums in any case
skipmd5sum="iffilled" # fill md5 sum only if they are not yet in db # default
#skipmd5sum="yes" # do not fill md5 sums in any case

doupdate="yes" # update all entries
doupdate="no" # fill only entries which are not yet existing #default

source `dirname $0`/../Sourcefile.sh
printprocesslog "INFO starting $0 with options doupdate="$doupdate" and skipmd5sum="$skipmd5sum

# setup to use ftools
source $HEADAS/headas-init.sh

# check if software is available
if ! ls $factpath/fitsdump >/dev/null 2>&1
then 
   printprocesslog "ERROR "$factpath"/fitsdump is not available."
   finish
fi

# check if paths are available
if ! ls /daq/raw >/dev/null 2>&1
then 
   printprocesslog "ERROR /daq/raw is not available."
   finish
fi
if ! ls /loc_data/raw >/dev/null 2>&1
then 
   printprocesslog "ERROR /loc_data/raw is not available."
   finish
fi
if ! ls /loc_data/zipraw >/dev/null 2>&1
then 
   printprocesslog "ERROR /loc_data/zipraw is not available."
   finish
fi

# get last 3, 6 or 9 nights
dates=( `date +%Y/%m/%d --date="-12hour"` `date +%Y/%m/%d --date="-36hour"` `date +%Y/%m/%d --date="-60hour"` \
#        `date +%Y/%m/%d --date="-84hour"` `date +%Y/%m/%d --date="-108hour"` `date +%Y/%m/%d --date="-132hour"` \
#        `date +%Y/%m/%d --date="-156hour"` `date +%Y/%m/%d --date="-180hour"` `date +%Y/%m/%d --date="-204hour"` \
        )
#dates=( `find /loc_data/zipraw/ -mindepth 3 -type d | sort | sed -e 's/\/loc_data\/zipraw\///g'` ) #all available dates in /loc_data/zipraw
#dates=( `find /loc_data/aux/ -mindepth 3 -type d | sort | sed -e 's/\/loc_data\/aux\///g'` ) #all available dates in /loc_data/zipraw

# do check for rawfiles of these dates
for date in ${dates[@]}
do 
   date2=`echo $date | sed -e 's/\///g'`
   night=
   query="SELECT fNight FROM AuxFilesAvailISDCStatus WHERE fNight="$date2
   night=`sendquery`
   if [ "$night" == "" ]
   then
      printprocesslog "INFO insert "$date2" to AuxFilesAvailISDCStatus"
      query="INSERT AuxFilesAvailISDCStatus SET fNight="$date2", fPriority="$date2
      sendquery >/dev/null 2>&1
   fi
   night=
   query="SELECT fNight FROM DriveFileAvailISDCStatus WHERE fNight="$date2
   night=`sendquery`
   if [ "$night" == "" ]
   then
      printprocesslog "INFO insert "$date2" to DriveFileAvailISDCStatus"
      query="INSERT DriveFileAvailISDCStatus SET fNight="$date2", fPriority="$date2
      sendquery >/dev/null 2>&1
   fi
   night=
   query="SELECT fNight FROM AuxDataInsertStatus WHERE fNight="$date2
   night=`sendquery`
   if [ "$night" == "" ]
   then
      printprocesslog "INFO insert "$date2" to AuxDataInsertStatus"
      query="INSERT AuxDataInsertStatus SET fNight="$date2", fPriority="$date2
      sendquery >/dev/null 2>&1
   fi
   ziprawdir=/loc_data/zipraw/$date
   # check if data are available from that night
   if ! [ -d $ziprawdir ]
   then
      printprocesslog "INFO "$ziprawdir" does not exist."
      continue
   else
      printprocesslog "INFO processing "$ziprawdir"..."
   fi

   # find all fits.gz files starting with the oldest file
   printprocesslog "INFO finding files to be checked in $ziprawdir..."
   fitsgzfiles=`find $ziprawdir -type f -name '*.fits.gz'| sort `

   # get runnumber from date
   runnumber=`echo $date | sed -e 's/\///g'`
   
   # loop to check files
   for file in $fitsgzfiles
   do
      printprocesslog "INFO checking file "$file
      #echo "INFO checking file "$file

      # raw and original file
      rawfile=`echo $file | sed -e 's/zipraw/raw/' -e 's/fits[.]gz/fits/'`
      origfile=`echo $rawfile | sed -e 's/loc_data/daq/'`

      # check if it is drs file
      isdrs=`ls $file | grep drs`
      if [ "$isdrs" != "" ]
      then
         printprocesslog "INFO "$file" is a drs file. -> continue"
         continue
      fi

      # check if file is already finished
      # original file on daq (if data was taken on daq
      if [ -e $origfile ]
      then
         # check if raw file was changed in the last 30 minutes
         isnew=`find $origfile -cmin -30`
         if [ "$isnew" != "" ]
         then
            printprocesslog "WARN "$origfile" is not older than 30 min. -> continue"
            continue
         fi

         # get time of last modification as seconds since Epoch for both files
         timeorig=`stat -c %Y $origfile`
         timecopy=`stat -c %Y $rawfile`
         # compare times
         if ! [ $timeorig -eq $timecopy ]
         then
            # if times are not the same, the file is still open => no check
            printprocesslog "INFO file "$rawfile" not yet closed. -> continue"
            continue
         fi
      else
         # if the origfile doesn't exist, the data was probably written not on daq but on data
         printprocesslog "INFO file "$rawfile" was probably taken on data and not daq."
      fi
      
      # get run and file number form filename
      runnumbererror="no"
      numbererror="no"
      numberfromname=`echo $file | grep -E -o '20[1-9][0-9][01][0-9][0-3][0-9]_[0-9]{3}'`
      runnumberfromname=`echo $numberfromname | cut -d_ -f1`
      filenumberfromname=`echo $numberfromname | cut -d_ -f2 | sed -e 's/^0//g' -e 's/^0//g'`
      if [ "$runnumber" != "$runnumberfromname" ]
      then
         runnumbererror="yes"
         printprocesslog "ERROR for file "$file": runnumber from date ("$runnumber") and filename ("$runnumberfromname") don't agree."
      fi
      
      # check if entry already exists
      query="SELECT fNight FROM RunInfo WHERE Concat(fNight, '_', LPAD(fRunID, 3, 0))='"$numberfromname"'"
      printprocesslog "DEBUG check if entry already exists in DB. QUERY: "$query
      #result3=`/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query3"`
      result3=`sendquery`

      # only proceed with file 
      #  if information is not yet in database
      #  and no update is wished ($doupdate)
      if [ "$result3" != "" ] && [ "$doupdate" == "no" ]
      then 
         printprocesslog "INFO "$file" has been inserted already. -> continue "
         continue
      fi
      
      runtype=
      # check if fits file is corrupted
      numfitserrors=0
      #checkfitsfile=`fverify $rawfile 2>/dev/null | grep '0 error(s)'`
      numfitserrors=`fverify $rawfile 2>/dev/null | grep -o '[0-9][ ]error(s)'  | grep -E -o '[0-9]'`
      #if [ "$checkfitsfile" == "" ]
      if [ $numfitserrors -gt 0 ]
      then
         printprocesslog "WARN "$rawfile" has "$numfitserrors" fitserror(s). "
         #numfitserrors=1
      fi

      numdrsfiles=
      step=
      if [ $numfitserrors -eq 0 ]
      then 
         # get run and file number from file
         runnumberfromfile=`$factpath/fitsdump -h -t Events $file 2>/dev/null | grep NIGHT | grep -E -o '20[1-9][0-9][01][0-9][0-3][0-9]'`
         filenumberfromfileorig=`$factpath/fitsdump -h -t Events $file 2>/dev/null | grep RUNID | grep -E -o '[0-9]{1,3}'`
         if [ "$runnumberfromfile" = "" ] || [ "$filenumberfromfileorig" = "" ]
         then
            printprocesslog "ERROR couldn't get run or file number from file name ("$file")."
         fi
         numberfromfile=$runnumberfromfile"_"`printf %03d $filenumberfromfileorig`
         # compare numbers
         if [ "$numberfromfile" != "$numberfromname" ]
         then
            numbererror="yes"
            printprocesslog "ERROR for file "$file": number from filename ("$numberfromname") and file ("$numberfromfile") don't agree."
         fi
      
         # check if this run has drs file
         #   in case file is available, get STEP from header
         # in the very beginning only drs-files were existing
         # in the beginning the keywords DRSCALIB and STEP were not existing
         drsfile=`echo $file | sed -e 's/fits/drs.fits/'`
         numdrsfiles=`ls $drsfile 2>/dev/null | wc -l`
         drscalib=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep DRSCALIB | grep -E -o "['][TF][']" | sed -e "s/'//g"`
         if [ "$drscalib" == "T" ]
         then 
            step=`$factpath/fitsdump -h -t Events $drsfile  2>/dev/null | grep STEP | grep -E -o "['][012][']" | sed -e "s/'//g"`
            if ! [ $numdrsfiles -eq 1 ]
            then 
               printprocesslog "ERROR for file "$file" number of drsfiles ("$numdrsfiles") and information from header ("$drscalib") don't agree."
            fi
         else
            if ! [ "$drscalib" == "F" ]
            then
               printprocesslog "WARN for file "$file" DRSCALIB is neither T nor F."
            fi
         fi
         
         # get other variables from header 
         runtype=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep RUNTYPE | grep -E -o "['][a-z0-9._-]+[']" | sed -e "s/'//g" -e "s/_/-/g" -e "s/[.]//g"`
         #echo "runtype for file "$file": "$runtype
         roi=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep NROI | grep -v NROITM | grep -E -o '[0-9]{1,4}'`
         roitm=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep NROITM | grep -E -o '[0-9]{1,4}'`
         numevents=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep Events | grep -E -o '[0-9]+'`
         numphys=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRG ' | grep -E -o '[0-9]+'`
         numext1=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGEXT1' | grep -E -o '[ ][0-9]+[ ]' | sed -e 's/\ //g'`
         numext2=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGEXT2' | grep -E -o '[ ][0-9]+[ ]' | sed -e 's/\ //g'`
         numelp=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGLPE' | grep -E -o '[0-9]+'`
         numilp=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGLPI' | grep -E -o '[0-9]+'`
         numoth=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGMISC' | grep -E -o '[0-9]+'`
         numped=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGPED' | grep -E -o '[0-9]+'`
         numtime=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'NTRGTIM' | grep -E -o '[0-9]+'`
         compiled=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'COMPILED' | grep -E -o "['][a-zA-Z]+[ ][ 12][0-9][ ]20[0-9][0-9][ ][0-2][0-9]:[0-5][0-9]:[0-5][0-9][']" | sed -e "s/'//g"`
         if ! [ "$compiled" == "" ]
         then 
            compiletime=`date +'%F %H:%M:%S' --date="${compiled}" `
         else
            compiletime=
         fi
         revnum=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep 'REVISION' | grep -E -o "['][0-9]+[:]?[0-9]*[MSP]*[']" | sed -e "s/'//g"`
         # in newest data start time is in DATE-OBS
         # in older data start time is in TSTART
         # in the beginning TSTART was empty
         runstart=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep DATE-OBS | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
         runstart2=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep TSTART | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
         if [ "$runstart" == ""  ]
         then
            if [ "$runstart2" == ""  ]
            then
               runstart=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep DATE | grep -v 'DATE-' | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
            else
               runstart=$runstart2
            fi
         fi
         # in newest data start time is in DATE-END
         # in older data start time is in TSTOP
         # in the beginning TSTOP was empty
         runstop=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep DATE-END | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
         runstop2=`$factpath/fitsdump -h -t Events $file  2>/dev/null | grep TSTOP | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
         if [ "$runstop" == ""  ]
         then
            if [ "$runstop2" == ""  ]
            then
               runstop=`stat $file  2>/dev/null | grep Modify | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9][ ][0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{9}'`
            else
               runstop=$runstop2
            fi
         fi
      fi
      
      # set runtype to 'unknown', if no runtype could be retrieved from file
      if [ "$runtype" == "" ]
      then
         runtype="n/a"
      fi
      # on 15.11.2011 the runtypes had different names
      if [ "$date" == "2011/11/15" ]
      then
         if [ "$runtype" == "drs-calib" ]
         then
            runtype="drs-gain"
         fi
         if [ "$runtype" == "drs-time-calib" ]
         then
            runtype="drs-time"
         fi
         if [ "$runtype" == "pedestal" ]
         then
            runtype="drs-pedestal"
         fi
         if [ "$runtype" == "light-pulser" ]
         then
            runtype="light-pulser-ext"
         fi
         if [ "$runtype" == "pedestal-on" ]
         then
            runtype="pedestal"
         fi
      fi
      # get runtype
      query="SELECT fRunTypeKEY FROM RunType WHERE fRunTypeName='"$runtype"'"
      printprocesslog "DEBUG get run type from DB. QUERY:"$query
      #result2=( `/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query2"` )
      result2=( `sendquery` )
      if [ ${#result2} -eq 0 ]
      then 
         printprocesslog "ERROR "$numberfromname": Could not query fRunTypeKey for runtype "$runtype" ."
         continue
      fi

      # check if entry has already checksums
      query="SELECT fNight FROM RunInfo WHERE Concat(fNight, '_', LPAD(fRunID, 3, 0))='"$numberfromname"'"
      query=$query" AND NOT ISNULL(fMd5sumRaw) AND NOT ISNULL(fMd5sumRawZip)"
      printprocesslog "DEBUG check if md5sums are alreay in DB. QUERY:"$query
      #result5=`/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query5"`
      result5=`sendquery`

      # get md5sums of raw and zip file
      #   to safe time for tests and update this can be skipped ($skipmd5sum)
      md5sum=
      md5sumzip=
      if [ "$skipmd5sum" == "no" ] || [ "$skipmd5sum" == "iffilled" ]
      then
         #echo "skip: "$skipmd5sum >> $logfile 2>&1
         #echo "res5: -"$result5"-" >> $logfile 2>&1
         #if [ "$skipmd5sum" == "no" ] || [ "$result5$skipmd5sum" == "iffilled" ] 
         if [ "$skipmd5sum" == "no" ] || [ "$result5" == "" ] 
         then 
            #zipfile=`echo $file | sed -e 's/raw/zipraw/' -e 's/fits/fits.gz/'`
            # only do the md5sum if the zipfile is already available
            if ls $file >/dev/null 2>&1
            then 
               printprocesslog "INFO calculation md5sum for rawfile "$rawfile
               md5sum=`md5sum $rawfile | cut -d' ' -f1`
               printprocesslog "INFO calculation md5sum for zipfile "$file
               md5sumzip=`md5sum $file | cut -d' ' -f1`
            fi
         fi
      fi

      # insert or update depending on whether run exists
      if [ "$result3" == "" ]
      then 
         query="INSERT"
         querymid=" fNight="$runnumber", fRunID="$filenumberfromname", "
         querystop=
      else
         query="UPDATE"
         querymid=
         querystop=" WHERE fNight="$runnumber" AND fRunID="$filenumberfromname
      fi
      query=$query" RunInfo SET "$querymid" fRunTypeKey="${result2[0]}
      if [ $numfitserrors -eq 0 ]
      then 
         query=$query", fRunStart='"$runstart"', fRunStop='"$runstop"'"
         if [ "$numevents" != "" ]
         then
            query=$query", fNumEvents="$numevents
         fi
         if [ "$roi" != "" ]
         then
            query=$query", fROI="$roi
         fi
         if [ "$roitm" != "" ]
         then
            query=$query", fROITimeMarker="$roitm
         fi
         if [ "$numphys" != "" ]
         then
            query=$query", fNumPhysicsTrigger="$numphys
         fi
         if [ "$numext1" != "" ]
         then
            query=$query", fNumExt1Trigger="$numext1
         fi
         if [ "$numext2" != "" ]
         then
            query=$query", fNumExt2Trigger="$numext2
         fi
         if [ "$numelp" != "" ]
         then
            query=$query", fNumELPTrigger="$numelp
         fi
         if [ "$numilp" != "" ]
         then
            query=$query", fNumILPTrigger="$numilp
         fi
         if [ "$numped" != "" ]
         then
            query=$query", fNumPedestalTrigger="$numped
         fi
         if [ "$numtime" != "" ]
         then
            query=$query", fNumTimeTrigger="$numtime
         fi
         if [ "$numoth" != "" ]
         then
            query=$query", fNumOtherTrigger="$numoth
         fi
      fi
      if [ "$md5sum" != "" ]
      then
         query=$query", fMd5sumRaw='"$md5sum"', fMd5sumRawZip='"$md5sumzip"'"
      fi
      if [ "$numdrsfiles" != "" ]
      then
         query=$query", fHasDrsFile="$numdrsfiles
      fi
      if [ "$step" != "" ]
      then
         query=$query", fDrsStep="$step
      fi
      if [ "$compiletime" != "" ]
      then
         query=$query", fCompileTime='"$compiletime"'"
      fi
      if [ "$revnum" != "" ]
      then
         query=$query", fRevisionNumber='"$revnum"'"
      fi
      query=$query", fFitsFileErrors="$numfitserrors
      query=$query" "$querystop
      # send query to DB
      sendquery >/dev/null
   done
done

finish