#! /bin/sh

# From the MTZ file which is the first file argument, create another with
# unique reflections and a column of free R flags.  Output is to file named 
# by the second file argument, else input with \`-unique' appended to its
# basename. Systematic absences are omitted from the output file unless
# the -s switch is specified. If free R flags are already present in
# the input file, use the -f <label> switch. If no free R flags are present
# then the fraction of reflections to be flagged with each indicator
# can be specified with the -p <fraction> switch (default 0.05).

# (This is Dave Love's interpretation of EJD's policy of adding
# reflections to complete the dataset generally, Nov. '95.  It depends
# on the current implementation of missing data values.)
# Modified by Martyn Winn to use cad rather than mtzutils. Added
# -s, -f <label> and -p <fraction> switches.

set -e

if test $# -lt 1 ; then
  cat 1>&2 <<+
Usage: `basename $0` [-s] [-f <label> | -p <fraction>] <input file>[.mtz] [<output file>]

 From the MTZ file which is the first file argument, create another with
 unique reflections and a column of free R flags.  Output is to file named 
 by the second file argument, else input with \`-unique' appended to its
 basename. Systematic absences are omitted from the output file unless
 the -s switch is specified. If free R flags are already present in
 the input file, use the -f <label> switch. If no free R flags are present
 then the fraction of reflections to be flagged with each indicator
 can be specified with the -p <fraction> switch (default 0.05).
+
  exit 1
fi

# read switches

param=true
conflict=false
SYSABS=" "
FREER=false
FREELABEL=" "
FRACT=0.05

while ($param); do

  case $1 in
#
# -s switch to keep systematic absences in mtz file
#
    "-s") SYSABS="SYSAB_KEEP"
          echo "Systematic absences will be kept." 1>&2
          shift ;;
#
# -f switch if freerflag column already present
#
     "-f") if ($conflict); then
              echo "-f switch incompatible with -p switch." 1>&2
              echo "Cannot specify FREERFRAC for existing Free-R column." 1>&2
              exit 1
           fi
           FREER=true
           conflict=true
           shift 
           FREELABEL=$1
           echo "Free-R flag column $FREELABEL is present." 1>&2
           shift ;;
#
# -p switch to specify FREERFRAC (new Free-R column only)
#
     "-p") if ($conflict); then
              echo "-p switch incompatible with -f switch." 1>&2
              echo "Cannot specify FREERFRAC for existing Free-R column." 1>&2
              exit 1
           fi
           conflict=true
           shift 
           FRACT=$1
           echo "Free-R flag column generated with FREERFRAC ${FRACT}." 1>&2
           shift ;;
#
# no more parameters
#
     *) param=false ;;
   esac

done

BASE=`basename $1 .mtz`

case $# in
1) OUT=$BASE-unique.mtz;logfile=$BASE-unique.log ;;
2) OUT=$2;logfile=`basename $2 .mtz`.log ;;
*) cat 1>&2 <<+
Wrong number of parameters!!

Usage: `basename $0` [-s] [-f <label> | -p <fraction>] <input file>[.mtz] [<output file>]
+
  exit 1;;
esac

logfiletmp=$logfile
i=0
while [ -f $logfiletmp ]
do
  i=`expr $i + 1`
  logfiletmp="${logfile}_${i}"
done
logfile=$logfiletmp
echo "Program output directed to $logfile" 1>&2

# Don't use $BASE since might have strange file suffix
if [ ! -r $1 ]; then
  if [ ! -r $1.mtz ]; then
    echo "$1 not found" 1>&2
    exit 1
  fi
fi

echo "Input file: $1 --- Output file: $OUT" 1>&2

temp1=$CCP4_SCR/uniq$$.mtz
temp2=$CCP4_SCR/uniq2$$.mtz
temp3=$CCP4_SCR/uniq3$$.mtz

# find awk executable
for myawk in nawk gawk awk
do
    test -f $myawk && break
done

# extract first crystal name from the input file
xname=`mtzdmp $1 -n 0 | $myawk '                  
/\* Dataset ID/ {
  getline; getline; getline;	# look 3 lines down
  print $1
  }'
`

# extract first dataset name from the input file
dname=`mtzdmp $1 -n 0 | $myawk '                  
/\* Dataset ID/ {
  getline; getline; getline; getline;	# look 4 lines down
  print $1
  }'
`

# From the input, extract what `unique' needs and then feed it:
mtzdmp $1 -n 0 |                # info from the input file
$myawk '                        # awk the right bits out
BEGIN { {ORS=""} }
# Process the file to get the information we need and write the
# the keywords
/\* Cell Dimensions :/ {  
  getline; getline;             # look 2 lines down
  print "CELL", $0, "\n"
  } 
/\* Space group = / {
  print "SYMMETRY"
  i = 5
  while ( $i !~/number/ ) {
    print "", $i
    i++
    }
  print "\n"
  }
/\*  Resolution Range :/ {
  getline; getline
  print "RESOLUTION", $6, "\n" # high res. only; most accurate in Aangstroems
  print "LABOUT F=FUNI SIGF=SIGFUNI\n"
  }
' |
unique hklout $temp1 > $logfile # with CELL, SYMMETRY, RESOLUTION, LABOUT already input

if ($FREER); then

#
#   Add this list to your existing F and SIGF and FREE
#
cad HKLIN2 $temp1 HKLIN1 $1 HKLOUT $temp2  << eof  >> $logfile
$SYSABS
LABI FILE 1  ALLIN
LABI FILE 2  E1=FUNI E2=SIGFUNI
XNAME FILE 2  E1=$xname E2=$xname
DNAME FILE 2  E1=$dname E2=$dname
END
eof

#
#  Extend FREE flags for extra hkl
#
freerflag HKLIN $temp2 HKLOUT $temp3 <<+  >> $logfile
COMPLETE FREE=$FREELABEL
END
+

#
#  mtzutils to get rid of useless UNIQUE labels, FUNI and SIGFUNI
#
mtzutils HKLIN $temp3 HKLOUT $OUT << eof >> $logfile
EXCLUDE FUNI SIGFUNI
eof

else

#
# Assign FreeR flags to these indices
# 
freerflag HKLIN $temp1 HKLOUT $temp2 <<+ >> $logfile
FREERFRAC $FRACT
END
+

#
# Use cad to append these FreeR flags to the real data.
#
cad HKLIN2 $temp2 HKLIN1 $1 HKLOUT $temp3 << + >> $logfile
$SYSABS
LABI FILE 1  ALLIN
LABI FILE 2  E1=FreeR_flag
XNAME FILE 2  E1=$xname
DNAME FILE 2  E1=$dname
END
+

#
# Ensure FreeR flags assigned to every reflection.
#
freerflag HKLIN $temp3 HKLOUT $OUT << + >> $logfile
COMPLETE FREE=FreeR_flag
END
+

fi

rm -f $temp1 $temp2 $temp3
#