#!/bin/bash

echo ""

SCRIPTNAME=$(basename $0)
HELP="Need help?  Try $SCRIPTNAME -h"
UERROR () {
echo -e "$1\n$HELP\n"
exit 1
} 

DOCDIR="$( cd "$( dirname "$0" )" && pwd )"/doc

# check optional inputs
while getopts "r:t:i:b:Rm:e:g:l:ph" OPTION; do
	case $OPTION in
	r)
		if [ $OPTARG -eq $OPTARG 2> /dev/null ]; then
			RUNDIR=$(printf 'r%04d' $OPTARG)
			RUNNUMBER=$(printf '%d' $OPTARG)
		else
			UERROR "Specify a numerical value for the run number"
		fi
		;;	
	t) TAG=$OPTARG;;
	l) UFILELIST=$OPTARG;;
	i) BOOST=$OPTARG;;
	b) BINNING=$OPTARG;;
	R) RANDOMIZE="YES";;
	m) MINPEAKS=$OPTARG;;
	e) RAWPATH=$OPTARG;;
	p) POWDER="POWDERS";;
	g) UGEOMFILE=$OPTARG
		[ -e $UGEOMFILE ] || UERROR "Geometry file $UGEOMFILE does not exist"
		;;
	h) SHOWHELP="YES";;
	?) UERROR "Unrecognized flag: $OPTION";;
	esac
done
shift `expr $OPTIND - 1`

if [ $SHOWHELP ]; then
cat << EOF
This is quickview, a handy (or just plain confusing?) bash script wrapper 
for Tom White's hdfsee program.

Rick Kirian, Oct. 2011

Brief summary of options:

   -h       Print this help message.
   -l xxx   Text file with a list of files to view.
   -r xxx   Run number.
   -t xxx   Tag name. 
   -m xxx   Skip images with fewer than this many peaks.
   -R       Randomly sort the images before viewing.
	-b xxx   Same as in hdfsee (set binning).
   -i xxx   Same as in hdfsee (set "boost").
   -e xxx   Same as in hdfsee (set hdf5 datapath).
   -g xxx   Same as in hdfsee (specify geometry file).
   -p	    Show powders: hits, then rejected frames.

This script is intended to be a handy way of passing lots of files to Tom
White's "hdfsee" program, a viewer for hdf5 files that have been created by 
Anton Barty's LCLS data analysis program "cheetah".  For starters, you should 
look at the documentation for hdfsee:

> hdfsee --help

Here are some examples of what quickview does.  In the simplest case, you can 
prepare file with a bunch of paths to hdffiles that you would like to view:

> quickview -l /some/path/list.txt

If you don't want to create the file, you can use quickview along with xargs
like this:

> find /some/path/to/hdf5/data -name LCLS*.h5 | xargs quickview

Now you can flip through images - if you close the current window, another
will open with the next file.  To exit, use control-c at the terminal (yes,
it's not very graceful).  

You might find the following: if there are peaks in the hdf5 file, they
will be overlaid on the image.  That is one purpose for the existence of
quickview: quickly check how peakfinding is going in cheetah.

For added convenience, quickview will attempt to look up an appropriate 
geometry file for you.  If you want this to happen, first define the
environment variable CONFIGDIR, which specifies the path where you will put
crystfel .geom files.  You'll need to make a geom file for each run you 
want to look at, with names like r0157.geom, and so on.  Most likely, you'll
want to create symbolic links, and have only a few .geom regular files.  Feel
free to put the geom files into a subdirectory, such as \$CONFIGDIR/geom - 
quickview searches recursively for these files.

For a bit more convenience, you can follow Rick's "tag" convention and give
your directories names like "r0234-rickstag".  Then you can run quickview like
this

> quickview -r 234 -t rickstag

to see the files in that directory.  This assumes that you have the environment
variable H5DIR set, which points to where these subdirectories are located.  

For a little bit more convenience, you can compare the results from two cheetah
sessions - try this:

> quickview -r 234 -t rickstag,nadiastag

In this case, quickview will show you all the files in the directory tagged with
nadiastag that are not in the one tagged with rickstag.  Sometimes it is nice to
see only the new results when changing the parameters in cheetah.

Some of the regular hdfsee flags are recognized by quickview.  Easy to add some
more if they would be helpful.

Finally, you can view only files with more than X peaks with the -m flag.  A
much better way to do this is to use quickview along with "framesorter" like
this

> framesorter [ OPTIONS ] | xargs quickview

Have a look at the framesorter help (i.e. type framesorter -h) for more info.

I think that's it.  If this all sounds crazy to you, that's because it is.  
Spending several months looking at hundreds of terabytes of crystallography
data does this to people.

One last thing: if you find annoying temporary files being written to your 
directories, set the environment variable \$TMPDIR - quickview will otherwise
guess where to write such files.

EOF
exit 0
fi

# defaults
[ -d $H5DIR ]    || H5DIR="."
[ -d $TMPDIR  ]  || TMPDIR="/tmp"
[ -d $TMPDIR ]   || TMPDIR="." #UERROR "The temp directory $TMPDIR does not exist"
[ $HDFSEE ]      || HDFSEE="hdfsee"
[ $H5DUMP ]      || H5DUMP="h5dump"
[ $BOOST ]       || BOOST="1"
[ $BINNING ]     || BINNING="2"
[ $RANDOMIZE ]   || RANDOMIZE="NO"
[ $MINPEAKS ]    || MINPEAKS="0"
[ $RAWPATH ]     || RAWPATH="/data/rawdata"
[ $POWDER ]	 || POWDER=""

# check the executables
[ -x $(which $HDFSEE) ]   || UERROR "Can't find executable $HDFSEE" 
[ -x $(which $H5DUMP) ]   || UERROR "Can't find executable $H5DUMP"

# set the hdfsee arguments
ARGS="$ARGS -e $RAWPATH -i $BOOST -b $BINNING"

# dump the list of files here
FILELIST=$TMPDIR/quickview.list
[ -f $FILELIST ] && rm $FILELIST

# if no filenames specified, then start looking for them
if [ $UFILELIST ]; then

	[ -f $UFILELIST ] || UERROR "Can't find file list $UFILELIST"
	cat $UFILELIST > $FILELIST
	NFILES=$(wc -l $FILELIST | awk '{print $1}')

elif [ $# -eq 0 ]; then

	[ $RUNNUMBER ] ||  UERROR "You did not specify a run number"
	echo "Run number: $RUNNUMBER"
	
	# sort out which save directories we're looking in
	SAVEDIR=$H5DIR/$RUNDIR
	COMPARETWO="NO"
	if [ $TAG ]; then
	
		# check if we are comparing two tags:
		[ $(echo $TAG | grep ',') ] && COMPARETWO="YES"
		TAG1=$(echo $TAG | awk -F ',' '{print $1}')
		TAG2=$(echo $TAG | awk -F ',' '{print $2}')
	
		if [ "$COMPARETWO" = "NO" ]; then
			SAVEDIR=$SAVEDIR-$TAG
			[ -d $SAVEDIR ] || UERROR "Data directory does not exist: $SAVEDIR"
		else
			if [ "$TAG1" = "" ]; then
				SAVEDIR1=$SAVEDIR
			else
				SAVEDIR1=$SAVEDIR-$TAG1
			fi	
			if [ "$TAG2" = "" ]; then
				SAVEDIR2=$SAVEDIR
			else
				SAVEDIR2=$SAVEDIR-$TAG2
			fi
			[ -d $SAVEDIR1 ] || UERROR "Data directory does not exist: $SAVEDIR1"
			[ -d $SAVEDIR2 ] || UERROR "Data directory does not exist: $SAVEDIR2"
		fi

	fi
		
	# write a file list with hdf5 file paths
	if [ "$COMPARETWO" == "NO" ]; then
		find $SAVEDIR -name 'LCLS*.h5' -type f > $FILELIST
		NFILES=$(wc -l $FILELIST | awk '{print $1}')
		echo "Data directory: $SAVEDIR ($NFILES files)"
	else
	
		# get the files in tag2, but not in tag1
		FILELIST1=$TMPDIR/quickview1.list
		FILELIST2=$TMPDIR/quickview2.list
		FILELIST3=$TMPDIR/quickview3.list
		FILELIST4=$TMPDIR/quickview4.list 
		find $SAVEDIR1 -name 'LCLS*.h5' | sed 's_.*/__'  | sort > $FILELIST1
		find $SAVEDIR2 -name 'LCLS*.h5' | sed 's_.*/__'  | sort > $FILELIST2
		cat $FILELIST1 $FILELIST2 | sort | uniq > $FILELIST3
		cat $FILELIST3 $FILELIST1 | sort | uniq -u > $FILELIST4
		rm $FILELIST &> /dev/null
		for i in $(cat $FILELIST4); do echo $SAVEDIR2/$i >> $FILELIST; done
	
		NFILES=$(wc -l $FILELIST | awk '{print $1}')
		echo "Data directory 1: $SAVEDIR1 ($(cat $FILELIST1 | wc -l) files)"
		echo "Data directory 2: $SAVEDIR2 ($(cat $FILELIST2 | wc -l) files)"
		[ $NFILES ] || NFILES="0"
		echo "Found $NFILES files in directory 2 that are not in directory 1"
	
	fi

        if [ "$POWDER" == "POWDERS" ]; then
                [ -f $FILELIST ] && rm $FILELIST
                FILELIST=$TMPDIR/quickviewpowder.list
                find $SAVEDIR -name 'r0*detector*-class*.h5' -type f > $FILELIST
                NFILES=$(wc -l $FILELIST | awk '{print $1}')
                echo "Data directory: $SAVEDIR ($NFILES virtual powder files)"
        fi

else

	# assume file paths have been provided as input
	for i in $@; do
		echo $i >> $FILELIST 
	done
	NFILES=$(wc -l $FILELIST | awk '{print $1}')

fi


# check that there are files to look at
if [ ! "$NFILES" -gt 0 ]; then
	echo -e "Nothing to look at\n"
	exit 0
else
	echo "Scanning through $NFILES files"
fi


# randomly sort the list, if requested
if [ "$RANDOMIZE" = "YES" ]; then

	echo "Randomly sorting file list"
	awk 'BEGIN {srand()} {print int(rand()*1000000) "\t" $0}' $FILELIST | \
		sort -n | cut -f 2- > $TMPDIR/quickview-randomized.list
	FILELIST=$TMPDIR/quickview-randomized.list

fi


# dump peaks to a text file, then have a look with hdfsee 
FILENUM=0
for FILE in $(cat $FILELIST); do

	FILENUM=$((FILENUM + 1))
	[ ! -f $FILE ] && continue
	echo $FILE > $TMPDIR/quickview-h5.list
	
	$H5DUMP -d /processing/hitfinder/peakinfo-raw $FILE | \
		awk '/^   \(/ {print $2 $3}' | awk -F ',' '{print $1, $2}' > \
		$TMPDIR/quickview-peaks.list
		
	NPEAKS=$(cat $TMPDIR/quickview-peaks.list | wc -l)

	printf 'File %d of %d: %s (%d peaks)\n' \
		$FILENUM $NFILES $(basename $FILE) $NPEAKS
	
	ARGSEXTRA=""
	if [ $UGEOMFILE ]; then
		ARGSEXTRA="$ARGSEXTRA -g $UGEOMFILE"
	else
		GEOMFILE=`expr match "$FILE" '.*\(r[0-9][0-9][0-9][0-9]\)'`.geom
		GEOMFILE=$(find $CONFIGDIR -name $GEOMFILE)
		[ $GEOMFILE ] && ARGSEXTRA="$ARGSEXTRA -g $GEOMFILE"
	fi 

	[ "$NPEAKS" -ge $MINPEAKS ] && \
		$HDFSEE $FILE $ARGS --show-rings $ARGSEXTRA -p $TMPDIR/quickview-peaks.list

done

echo "Done"
echo ""