#!/bin/bash

############################################################################
#
# Version: $Revision: 1.2 $
#
# Date: $Date: 2004/09/15 23:23:53 $
#
# Author	: Cody Green
# 		  Digital Initiatives Research
#		  Texas A&M University
# Email 	: codygreen@tamu.edu
# Description	: Shell script that helps migrate data from a development
# 		  server to a production server.  It will remove the handle 
#		  and any entries in the dublin_core.xml file that will be 
#		  duplicated upon importing the data into DSpace.
#
############################################################################

# Make sure user is in bash
if [ -z "$BASH" ]; then
	printf "ERROR: please run this script with the BASH shell.\n" 
	exit 192
fi

# Check if SED is in the path
which sed > /dev/null
if [ $? -ne 0 ]; then      
	# Check what OS user has
	if [ `uname -s` = "SunOS" ]; then 
		# Declare variables
		declare -rx SED="/usr/bin/sed"
	fi
	
	if [ `uname -s` = "Linux" ]; then
		# Declare variables
		declare -rx SED="/bin/sed"
		fi
	
	# Sanity check for programs
	if [ ! -e $sed ]; then
		printf "ERROR: Can not find the program sed.\n 
		please check that the program is installed or 
		in your path"
		exit 1
	fi
else
	export SED=`which sed`
fi

# Check parameter count, show usage if incorrect number passed
if [ $# -eq 0 ]; then
        printf "Usage: dspace_migrate.sh [DIRECTORY]\n"
        printf "Example: /dspace/bin/dspace_migrate.sh /home/user/ETDs/ \n\n"
        printf "This script will remove the handle and any \nentries in the dublin_core.xml file that will \nbe duplicated upon importing the data into DSpace.\n\n"
        exit 0;
fi

# Check if users needs help
if [ $1 = "--help"  -o $1 = "--h" ]; then
	printf "Usage: dspace_migrate.sh [DIRECTORY]\n"
	printf "Example: /dspace/bin/dspace_migrate.sh /home/user/ETDs/ \n\n"
	printf "This script will remove the handle and any \nentries in the dublin_core.xml file that will \nbe duplicated upon importing the data into DSpace.\n\n"
	exit 0;
fi

# Check to see if the directory exists
if [ ! -d $1 ]; then
	printf "ERROR: $1 is not a valid directory.\n"
	exit 1;
fi

printf "Fixing Bad Data in ETDs\n"

# Loop through the directory
for i in ls ./$1/*/dublin_core.xml; 
do 
	 # Ignore ls in the returned values
   if [ $i = "ls" ]; then
      continue
   fi	
   printf "Checking $i...\n"
   # Check if file exists
   if [ ! -s $i ]; then
      printf "ERROR: $i does not exist of is empty\n"
      exit 1
   fi 
   # copy file
   cp $i $i.orig
   if [ $? -ne 0 ]; then
      printf "ERROR: Could not copy $i to $i.orig \n"
      exit 1
   fi

   printf "   --removing null and duplicate values for $i\n"
   $SED "/><\/dcvalue>/d" $i | $SED "/element=\"date\" qualifier=\"accessioned\"/d" | $SED "/element=\"date\" qualifier=\"available\"/d" | $SED "/element=\"date\" qualifier=\"issued\"/d" | $SED "/element=\"description\" qualifier=\"provenance\"/d" | $SED "/element=\"identifier\" qualifier=\"uri\"/d" | $SED "/element=\"format\" qualifier=\"extent\"/d" | $SED "/element=\"format\" qualifier=\"mimetype\"/d" | $SED "/bytes, checksum/d" > $i

   if [ $? -ne 0 ]; then
      printf "ERROR: Could not fix $i\n"
      exit 1
   fi

   # Check to see if filesize is 0
   printf "   --checking filesize for $i\n"
   if [ ! -s $i ]; then
      printf "ERROR: filesize for $i is 0\n"
      exit 1
   fi

   # remove copy of file
   rm -f $i.orig
   if [ $? -ne 0 ]; then
      printf "ERROR: Could not remove $i.orig\n"
      exit 1
   fi

done
rm -f ./$1/*/handle
if [ $? -ne 0 ]; then
   printf "ERROR: Could not delete handles\n"
   exit 1
fi
printf "Data Has Been Fixed\n"