#!/usr/bin/ksh

#
# Directory File analysis.
# Ben Mealey - June 2001
#

# Format field widths.

typeset -R15  allFiles
typeset -R15  allDirs
typeset -R15  allLinks

typeset -R15  Mtotal
typeset -R15  Ktotal
typeset -R15  total

typeset -R15  count
typeset -R15  AFcount

typeset -L5   pfileType

if (( $# != 1 )); then
  print "Usage: $0 <dir>"
  exit
fi

if [[ ! -d $1 ]] then
  print "Eh? - $1 not a directory"
  exit
fi

parmDir=$1

(( allFiles = 0 ))
(( allDirs  = 0 ))
(( allLinks = 0 ))

(( count = 0 ))
(( AFcount = 0 ))

print "$(uname -n) - Files under ${parmDir} - $(date)"

#
# File type analysis
#
# The first find/grep generates a list of files that end in the 
# familiar filetype format, for example "sample.gif".
#                                              XXXX
#
# The awk -F | sort | uniq | sort -rn produces a sorted list of all
# filetypes.  For example:
#
#            15 gif
#            12 tif
# 
# and so on.
#
# Each filetype is fed into a second find to calculate the size, in
# bytes of matching files.
#
# Compare the "Total analyzed files" with the subsequent "Total Files'
# in order to get a count of files which didn't match the classic
# filename.fileExtension pattern.
#

print " "
print "File type count:"
print " "
print "  Type           Count            Size"

find ${parmDir} -type "f" -print | grep "\.[^/]*" | awk -F"." '{print $NF}' | sort | uniq -c | sort -rn | while read line; do
  (( count = 0 ))
  (( total = 0 ))
  fileType="*$(print ${line} | awk '{print $2}')"

  find ${parmDir} -name "${fileType}" -type "f" -print -xdev |\
   xargs -I{} ls -l {} |\
   awk '{print $5 " " $NF}' | while read type; do
     size="$(print ${type} | awk '{print $1}')"
     fn1="$(print ${type} | awk '{print $2}')"
     fn="${fn1##*/}"
     (( count = count + 1 ))
     (( total = total + size ))
   done

#
# Scale raw byte counts into either MB or K.
#

   pfileType=${fileType}  # padded version

   if (( total > 1000000 )); then
     (( Mtotal = total / 1000000 ))
     print "${pfileType} ${count} ${Mtotal} MB"
   elif (( total > 1000 )); then
     (( Ktotal = total / 1000 ))
     print "${pfileType} ${count} ${Ktotal} K Bytes"
   else
     print "${pfileType} ${count} ${total} Bytes"
   fi

  (( AFcount = AFcount + count ))

done

print " "
print "Total analyzed files: ${AFcount}"
print " "

#
# Get total count of files, directories, and links.
#
# Size discrepancies are due to rounding errors, and the different
# ways that the ls -l and du different commands report space. 
#

(( allFiles = $(find ${parmDir} -type "f" -print | wc -l) ))
(( allDirs  = $(find ${parmDir} -type "d" -print | wc -l) ))
(( allLinks = $(find ${parmDir} -type "l" -print | wc -l) ))

print "Total Files           ${allFiles}"
print "Total Directories     ${allDirs}"
print "Total Links           ${allLinks}"

#
# Get space used.
#

print " "

(( KspaceUsed = $(du -ks ${parmDir} | awk '{print $1}') )) 

# Report in MB if space is more than 10,000 K or 10 MB.

if (( KspaceUsed > 10000 )); then
  (( MspaceUsed = KspaceUsed / 1000 ))
  print "Space used:                        ${MspaceUsed} MB"
else
  print "Space used:                        ${KspaceUsed} K"
fi
print " "

print "Done  - $(date)"