akubra redist

gistfile1.txt

#!/bin/bash

#note this program requires the setting of three variables:
# 1) the original akubra directory
# 2) the destination akubra directory
# 3) and awkcmd representing the hash structure

time_start=$(date)

orig_dir='/usr/local/fed362/data/objectStore'
#orig_dir='/usr/local/fed362/data/datastreamStore'

dest_dir='/usr/local/fed362/data/objectStore_dest2'
#dest_dir='/usr/local/fed362/data/datastreamStore_dest'

file_count = 0
dir_count = 0

##
##/##  = 16,777,216
##/##/# = 268,435,456

#awkcmd='{ print substr($0,1,2); }'
awkcmd='{ print substr($0,1,2) "/" substr($0,3,2); }'
#awkcmd='{ print substr($0,1,2) "/" substr($0,3,2) "/" substr($0,5,1); }'

echo "Origin dir:" $orig_dir
echo "destination dir:" $dest_dir

function directory_recurse() {
  for ff in "$1"/*
  do
    if [ -d "$ff" ]
	then
	  dir_count=$[$dir_count +1]
	  #echo "directory $ff"
      directory_recurse $ff		
	elif [ -f "$ff" ]
 	then
	  file_count=$[$file_count +1]
      #echo "oldfilepath $ff"
	  filename=$(basename $ff)
	  path=`echo -n $filename | sed 's/%3A/:/g' | sed -e 's=%2F=/=g' | md5sum | awk "$awkcmd"`
	  fullpath=$dest_dir/$path/$filename
	  mkdir -p $dest_dir/$path
	  echo "cp $ff $dest_dir/$path/$filename"
      cp $ff $dest_dir/$path/$filename	  
    fi
  done 	  
}

for f in $orig_dir
do
  if [ -d "$f" ]
  then
    directory_recurse $f  
  fi
done
echo "---"
echo  "file count:" $file_count
echo "dir count:" $dir_count

time_end=$(date)
echo "start:" $time_start
echo "end:" $time_end