#!/bin/bash

if [ ! -e /usr/bin/awk ]; then 
  echo Please make sure you have installed awk!
  exit 1
fi

if [ $# -lt 2 ] ; then
        
        cat <<EOFHELP

        "musrfit-mlog to db converter"

USAGE:   mlog2db FIRSTRUN# LASTRUN# EXTENSION [noheader | nosummary | fit-TEMPLATERUN# | -oDB_OUTPUT_FILE | -k]
         OR
         mlog2db \[SPACE SEPARATED LIST OF RUNS\] EXTENSION [noheader | nosummary | fit-TEMPLATERUN# | -oDB_OUTPUT_FILE | -k]
         OR
         mlog2db RUNLISTFILE EXTENSION [noheader | nosummary | fit-TEMPLATERUN# | -oDB_OUTPUT_FILE | -k]

This small script converts subsequent musrfit output files (mlog) into one db file using bash/awk.

Example 1:
mlog2db 1423 1425 _tf_h13
generates the db-file "out.db" (or whatever filename you additionally specify with the -o option) from 1423_tf_h13.mlog, 1424_tf_h13.mlog, and 1425_tf_h13.mlog .

Example 2:
mlog2db [1487 1435] _tf_h13
generates the db-file "out.db" (or whatever filename you additionally specify with the -o option) from 1487_tf_h13.mlog and 1435_tf_h13.mlog .

Example 3:
mlog2db runs.txt _tf_h13
generates the db-file "out.db" (or whatever filename you additionally specify with the -o option) from all runs listed in the ASCII-file
"runs.txt" in the working directory. In the file it's also possible to include external parameters which should be put in the resulting db-file.
The structure of the "runs.txt" is the following:

RUN VAR1 VAR2 VAR3 ...
2716 200 27.1 46.2 ...
2787 205 27.1 46.3 ...
2260 210 27.2 45.9 ...
...

The first line specifies the db-parameter names and labels and has to be present!

The output files in all examples above also include the db header.
In the case you are dealing with LEM-data AND have locally mounted the /mnt/data/nemu directory some available parameters like the 
temperature or transport settings will be taken from the summary files and added to the db-file.
You can suppress the reading of the summary files with the option "nosummary".

If you want to generate a file without the header information just run the script with the option "noheader".

If you additionally want to fit some data using musrfit specify the option "fit-TEMPLATERUN#".
In this case either a mlog-outputfile or a msr-inputfile have to be present for the specified template run.
The subsequent inputfiles will be created.

Example 4:
mlog2db 200 220 _tf_h13 -oABC.db fit-199

This will fit the runs 200 to 220 using musrfit. The file 199_tf_h13.mlog (or if that is not available: 199_tf_h13.msr) is used as 
template for the first musrfit input file. The results of the fits will be written to ABC.db.

For keeping the MINUIT2 output files after calling musrfit, i.e. invoke musrfit with the option "--keep-mn2-output" just pass the option "-k" to mlog2db. If you pass this option to mlog2db but do not fit any data, the option will just be ignored.

If your .mlog files do not have an extension as specified above like "200.mlog" you have to call the script as in the following example.

Example 5:
mlog2db 200 220 "" -oABC.db fit-199

CAUTION:
The "indexing number" of the .mlog has to be at the begin of the filename.
Furthermore the data files that were fitted have to have the name "XXX_RUN#[_YYY]", where XXX,YYY are strings without whitespaces ("/" is OK).

If you want to use the fitting feature, musrfit has to be installed either on the PATH or in the working directory.

Those of you who do not like scripting without functions better do not look at the code. Still it is more or less straight forward.

EOFHELP

else

  export PATH=./:$PATH

  SUMMDIR="/mnt/data/nemu/summ"

  # In case a list of runs is given by [...]
  if [ "${1:0:1}" == "[" ]; then
    RUNARRAY=($@)
    if [ "$1" == "[" ]; then
      RUNARRAY[0]=""
    else
      RUNARRAY[0]=${RUNARRAY[0]#"["}
    fi
    tLen=${#RUNARRAY[@]}
    for (( i=0; i<${tLen}; i++ ))
    do
      if [ "$1" == "[" ]; then
        RUNARRAY[$i]=${RUNARRAY[ (( i + 1 )) ]}
      fi
      if [ "${RUNARRAY[$i]:(-1)}" == "]" ]; then
        if [ "${RUNARRAY[$i]}" == "]" ]; then
          RUNARRAY[$i]=""
          LASTRUN=$(( i - 1 )) # RUNARRAY-index of the last run
        else
          RUNARRAY[$i]=${RUNARRAY[$i]%"]"}
          LASTRUN=$i # RUNARRAY-index of the last run
        fi
      fi
    done

    # search for the parameter including the "]"
    for (( i=0; i<="$#"; i++ ))
    do
      RIGHTBRACKET=$( eval echo \$$i )
      if [ "${RIGHTBRACKET:(-1)}" == "]" ]; then
        RIGHTBRACKET=$i
        break
      fi
    done

    EXTENSION=$( eval echo \$$((RIGHTBRACKET + 1)) )
    OPT_PAR_ONE=$( eval echo \$$((RIGHTBRACKET + 2)) )
    OPT_PAR_TWO=$( eval echo \$$((RIGHTBRACKET + 3)) )
    OPT_PAR_THREE=$( eval echo \$$((RIGHTBRACKET + 4)) )
    OPT_PAR_FOUR=$( eval echo \$$((RIGHTBRACKET + 5)) )
    OPT_PAR_FIVE=$( eval echo \$$((RIGHTBRACKET + 6)) )
    RUNLIST_USED=0
    NUM_OF_IND_VAR=1 # 1 for the run number

  else 
    ISDIGIT=$( echo $1 | awk '{ if (( $1 + 0 ) == $1 ) {print 1} else {print 0}}' ) # check if the first parameter is a number or not
    if [ "$ISDIGIT" -eq 1 ]; then # start and end-runs are given
      for (( j=0; j<=$2-$1; j++ ));
      do
        RUNARRAY[$j]=$(( $1 + $j ))
      done
      LASTRUN=$(( $2 - $1 )) # RUNARRAY-index of the last run
      EXTENSION=$3

      OPT_PAR_ONE=$4
      OPT_PAR_TWO=$5
      OPT_PAR_THREE=$6
      OPT_PAR_FOUR=$7
      OPT_PAR_FIVE=$8
      RUNLIST_USED=0
      NUM_OF_IND_VAR=1 # 1 for the run number

    else # assume a runlist-file with "independent variables" is given
      if [ ! -e $1 ]; then
        echo
        echo "The specified runlist file" $1 "is not present... Bye bye!"
        echo
        exit 1
      fi

      numberOfReadInLines=0

      exec 3< $1 # open file for reading in the runlist-file the first time and get the runnumbers to be processed
      while read -u 3 -a runList
      do
        if [ "$numberOfReadInLines" == "0" ]; then
          NUM_OF_IND_VAR=${#runList[@]}
        else
          RUNARRAY[(( numberOfReadInLines - 1 ))]=${runList[0]}
        fi
        ((numberOfReadInLines++))
      done
      exec 3<&- # close file

      LASTRUN=$(( ${#RUNARRAY[@]} - 1 )) # RUNARRAY-index of the last run
      EXTENSION=$2

      OPT_PAR_ONE=$3
      OPT_PAR_TWO=$4
      OPT_PAR_THREE=$5
      OPT_PAR_FOUR=$6
      OPT_PAR_FIVE=$7
      RUNLIST_USED=1
    fi
  fi

  # Check if fitting should be done
  if [ "${OPT_PAR_ONE:0:3}" == "fit" ]; then
    TEMP=${OPT_PAR_ONE:4}
  elif [ "${OPT_PAR_TWO:0:3}" == "fit" ]; then
    TEMP=${OPT_PAR_TWO:4}
  elif [ "${OPT_PAR_THREE:0:3}" == "fit" ]; then
    TEMP=${OPT_PAR_THREE:4}
  elif [ "${OPT_PAR_FOUR:0:3}" == "fit" ]; then
    TEMP=${OPT_PAR_FOUR:4}
  elif [ "${OPT_PAR_FIVE:0:3}" == "fit" ]; then
    TEMP=${OPT_PAR_FIVE:4}
  fi

  if [ "$TEMP" != "" ]; then # runs should be fitted using musrfit

  # Check for musrfit on the PATH and if found continue with searching the fit-template
  PATHTOMUSRFIT=$(echo `which musrfit`)
    if [ "$PATHTOMUSRFIT" == "" ]; then
      echo
      echo "No musrfit executable was found on the PATH or in the current directory!"
      echo "Please install musrfit first!"
      echo
      exit 1
    else
      if [ -e $TEMP$EXTENSION.mlog ]; then
        TEMPLATE=$TEMP$EXTENSION.mlog
      elif [ -e $TEMP$3.msr ]; then
        TEMPLATE=$TEMP$EXTENSION.msr
      fi
    fi
  fi

  for (( q=0; q<=LASTRUN; q++ )) # loop over all RUNARRAY entries - the actual data-processing
  do
    COUNT=${RUNARRAY[$q]}
    NEXTCOUNT=${RUNARRAY[ (( q + 1 )) ]}
    NEXTINPUT=$NEXTCOUNT$EXTENSION.msr
    FIRSTINPUT=${RUNARRAY[0]}$EXTENSION.msr

    if [ "$TEMP" != "" ]; then # if fitting should be done
      if [ "$TEMPLATE" == "" ]; then
        echo
        echo "The specified musrfit template files "$TEMP$3".mlog and $TEMP$3.msr do not exist!"
        echo
        exit 1
      fi

      # if it's the first run to be fitted, take the template and substitute the runnumber to match the first run
      if [ "$COUNT" == "${RUNARRAY[0]}" ]; then
        awk -v count=$TEMP -v nextcount=${RUNARRAY[0]} -v nextinput=$FIRSTINPUT '{
          if(NR==1) print nextcount > nextinput
          else if($1 == "RUN") {
            if (substr(count,1,1) != "_"){ count = sprintf("_%04u", count); nextcount = sprintf("_%04u", nextcount);}
            sub(count, nextcount, $0);
            print $0 >> nextinput
          }
          else print $0 >> nextinput
        }' $TEMPLATE
      fi

      if [ "${OPT_PAR_ONE}" == "-k" ] || [ "${OPT_PAR_TWO}" == "-k" ] || [ "${OPT_PAR_THREE}" == "-k" ] || [ "${OPT_PAR_FOUR}" == "-k" ] \
      || [ "${OPT_PAR_FIVE}" == "-k" ]; then
        MUSRFITPARAM="-k"
      fi

      musrfit $COUNT$EXTENSION.msr $MUSRFITPARAM # the fit

      if [ "$COUNT" != "${RUNARRAY[$LASTRUN]}" ]; then # create the next musrfit input file if not the last specified run was fitted already
        awk -v count=$COUNT -v nextcount=$NEXTCOUNT -v nextinput=$NEXTINPUT '{
          if(NR==1) print nextcount > nextinput
          else if($1 == "RUN") {
            if (substr(count,1,1) != "_"){ count = sprintf("_%04u", count); nextcount = sprintf("_%04u", nextcount);}
            sub(count, nextcount, $0);
            print $0 >> nextinput
          }
          else print $0 >> nextinput
        }' $COUNT$EXTENSION.mlog
      fi
    fi

    if [ ! -e $COUNT$EXTENSION.mlog ]; then
      echo
      echo "The specified musrfit output file "$COUNT$EXTENSION".mlog does not exist!"
      echo
      exit 1
    fi

    # if the "nosummary"-option is not set and the summary files are present, get some information from there
    if [ "${OPT_PAR_ONE}" != "nosummary" ] && [ "${OPT_PAR_TWO}" != "nosummary" ] && [ "${OPT_PAR_THREE}" != "nosummary" ] \
    && [ "${OPT_PAR_FOUR}" != "nosummary" ] && [ "${OPT_PAR_FIVE}" != "nosummary" ]; then

      if [ -d $SUMMDIR ]; then
        # extract runnumber from mlog-file
        RUNNUMBER=$( awk '{
          if(NR==1) X=0
          if($1 ~ /^RUN/ && X==0) {
            split($2, runNumber, "_")
            if(runNumber[1] ~ /lem/ && runNumber[2] != "his") {print runNumber[2] }
            if(runNumber[1] ~ /lem/ && runNumber[2] == "his") {print substr(runNumber[3],1,4) }
            X=1 
          }
        }' $COUNT$EXTENSION.mlog )

        # extract runyear from mlog-file in case of LEM-data
        RUNYEAR=$( awk '{
          if(NR==1) X=0
          if($1 ~ /^RUN/ && X==0) {
            split($2, runNumber, "_")
            if(runNumber[1] ~ /lem/) {print substr(runNumber[1],length(runNumber[1])-1)}
            else print "XX"
            X=1
          }
        }' $COUNT$EXTENSION.mlog )

        if [ "$RUNYEAR" != "XX" ]; then # in case of LEM-data search the summary-file
          SUMMFILE=$SUMMDIR/20$RUNYEAR/lem$RUNYEAR\_$RUNNUMBER.summ
          TITLE=$( awk '{ if(NR==4) print $0 }' $SUMMFILE )
          ENERGY=$( awk '{ if($0 ~ /implantation energy/) print $(NF-1)}' $SUMMFILE ) 
          TEMPERATURE=$( awk '{ if($1 == "Sample_CF1") print $(NF-1)}' $SUMMFILE )
          RALRAR=$( awk '{ if($5 == "RA-L") RAL=$7; if($1 == "RA-R") { RAR=$3; print RAL-RAR; nextfile } }' $SUMMFILE )
          RATRAB=$( awk '{ if($5 == "RA-T") RAT=$7; if($1 == "RA-B") { RAB=$3; print RAT-RAB; nextfile } }' $SUMMFILE )
          TRANSPORT=$( awk '{ if($1 == "Moderator") print $3}' $SUMMFILE )
        fi
      fi
    fi

    # put together all data to db-format and write it to the file
    awk -v optPARone=$OPT_PAR_ONE -v optPARtwo=$OPT_PAR_TWO -v optPARthree=$OPT_PAR_THREE -v optPARfour=$OPT_PAR_FOUR \
    -v optPARfour=$OPT_PAR_FIVE -v runList=$RUNLIST_USED -v numIndVar=$NUM_OF_IND_VAR '{

      outfile="out.db"
      if(optPARone ~ /^-o/) outfile = substr(optPARone,3)
      if(optPARtwo ~ /^-o/) outfile = substr(optPARtwo,3)
      if(optPARthree ~ /^-o/) outfile = substr(optPARthree,3)
      if(optPARfour ~ /^-o/) outfile = substr(optPARfour,3)
      if(optPARfive ~ /^-o/) outfile = substr(optPARfive,3)

      FS = " "
      OFS = " "

      if(NR==1){title=$0; X=0; Y=0; i=1}

      if(NR>4){
        if(X == 0){
          if($0 != ""){
            parArray[i] = $2
            dataArray[i] = $3
            if($4 ~ /^-/)
              negErrArray[i] = (-1.0)*$4
            else
              negErrArray[i] = $4
            if($5 ~ /none/)
              posErrArray[i] = $4
            else
              posErrArray[i] = $5
            i++
          }
          if($0 == "") X=1
        }
      }

      if($1 ~ /^RUN/ && X==1){ split($2, runNumber, "_"); X=2 }	

      if($1 ~ /^chi/ && X==2){
        parArray[i] = "CHISQ"
        parArray[i+1] = "NDF"
        parArray[i+2] = "CHISQred"
        parArray[i+3] = "RUN"

        sub(/,/, "", $3); sub(/,/, "", $6);

        dataArray[i] = $3
        dataArray[i+1] = $6
        dataArray[i+2] = $9

        for (j in runNumber) {
          if ( runNumber[j] ~ /^0/ ) {
            sub(/0+/, "", runNumber[j])
          }
          if (int(runNumber[j]) == runNumber[j]) {
            dataArray[i+3] = runNumber[j]
            break
          }
          else dataArray[i+3] = "0000"
        }

        negErrArray[i] = ""
        negErrArray[i+1] = ""
        negErrArray[i+2] = ""
        negErrArray[i+3] = ""

        posErrArray[i] = ""
        posErrArray[i+1] = ""
        posErrArray[i+2] = ""
        posErrArray[i+3] = ""	

        # Read in the runlist-variable names
        if(runList == 1){
          getline line < "'"$1"'"
          split(line, indVar, " ")
        }

        # Output to file in the case, the LEM summary file is accessible and the option "nosummary" is not set

        if("'"$TEMPERATURE"'" != "") {

          # Write the db-header
          if("'"$COUNT"'" == "'"${RUNARRAY[0]}"'" && optPARone != "noheader" && optPARtwo != "noheader" && optPARthree != "noheader" && optPARfour != "noheader" && optPARfive != "noheader"){

            print "TITLE" >> outfile
            print ">>>Put your title here<<<\n" >> outfile
            print "Abstract" >> outfile
            print ">>>Put your abstract here<<<\n" >> outfile
            print "LABELS" >> outfile
            print "T [K]" >> outfile
            print "Tr [kV]" >> outfile
            print "E [keV]" >> outfile
            print "RAL-RAR [kV]" >> outfile
            print "RAT-RAB [kV]" >> outfile

            if(runList == 1){
              for(k=2;k<=numIndVar;k++){print indVar[k] >> outfile}
            }

            for(k=1;k<i+4;k++){
              print parArray[k] >> outfile
            }

            $1 = "Data"
            $2 = "T"
            $3 = "Tr"
            $4 = "E"
            $5 = "RALRAR"
            $6 = "RATRAB"

            for(k=1;k<numIndVar;k++){
              $(k+6)=indVar[k+1]
            }

            for(k=1;k<i+4;k++){
              $(k+6+numIndVar-1)=parArray[k]
            }

            print "\n" $0 >> outfile
            print "\\-e" >> outfile
          }

          # Write the data-block
          print "T = " "'"$TEMPERATURE"'" ",,,\\" >> outfile
          print "Tr = " "'"$TRANSPORT"'" ",,,\\" >> outfile
          if ("'"$ENERGY"'" != "") print "E = " "'"$ENERGY"'" ",,,\\" >> outfile
          else print "E = -999,,,\\" >> outfile
          if ("'"$RALRAR"'" != "") print "RALRAR = " "'"$RALRAR"'" ",,,\\" >> outfile	
          else print "RALRAR = -999,,,\\" >> outfile
          if ("'"$RATRAB"'" != "") print "RATRAB = " "'"$RATRAB"'" ",,,\\" >> outfile
          else print "RATRAB = -999,,,\\" >> outfile

          if(runList == 1) { # process the "independent variables" from the file
            while((getline line < "'"$1"'") > 0){
              split(line, indVarValue, " ")
              if (indVarValue[1] == "'"$COUNT"'"){
                for(k=2;k<=numIndVar;k++){print indVar[k] " = " indVarValue[k] ",,,\\" >> outfile}
                break
              }
            }
          }

          for(l=1;l<i+3;l++){
            print parArray[l] " = " dataArray[l] ", " posErrArray[l] ", " negErrArray[l] ",\\" >> outfile
          }

          print dataArray[i+3] ",,, " "'"$TITLE"'" >> outfile

          if("'"$COUNT"'" == "'"${RUNARRAY[$LASTRUN]}"'") print "\n" >> outfile
        }

        # If we do not have the information from the summary file or "nosummary" is set
        else {
          if("'"$COUNT"'" == "'"${RUNARRAY[0]}"'" && optPARone != "noheader" && optPARtwo != "noheader" && optPARthree != "noheader" && optPARfour != "noheader" && optPARfive != "noheader"){

            print "TITLE" >> outfile
            print ">>>Put your title here<<<\n" >> outfile
            print "Abstract" >> outfile
            print ">>>Put your abstract here<<<\n" >> outfile
            print "LABELS" >> outfile

            if(runList == 1){
              for(k=2;k<=numIndVar;k++){print indVar[k] >> outfile}
            }

            for(k=1;k<i+4;k++){
              print parArray[k] >> outfile
            }

            $1 = "Data"

            for(k=1;k<numIndVar;k++){
              $(k+1)=indVar[k+1]
            }

            for(k=1;k<i+4;k++){
              $(k+1+numIndVar-1)=parArray[k]
            }

            print "\n" $0 >> outfile
            print "\\-e" >> outfile
          }

          if(runList == 1) {
            while((getline line < "'"$1"'") > 0){
              split(line, indVarValue, " ")
              if (indVarValue[1] == "'"$COUNT"'"){
                for(k=2;k<=numIndVar;k++){print indVar[k] " = " indVarValue[k] ",,,\\" >> outfile}
                break
              }
            }
          }

          for(l=1;l<i+3;l++){
            print parArray[l] " = " dataArray[l] ", " posErrArray[l] ", " negErrArray[l] ",\\" >> outfile
          }

          print dataArray[i+3] ",,, " title >> outfile

          if("'"$COUNT"'" == "'"${RUNARRAY[$LASTRUN]}"'") print "\n" >> outfile
        }
      }
    }' $COUNT$EXTENSION.mlog
  done
fi