#!/bin/ksh
##
## SCRIPT NAME:  find_files4filetype
##
## Where: in $FEDIR/scripts   where $FEDIR=/apps/nns_com/fea
##
##############################################################################
##
## PURPOSE: For a specified directory residing-on (or known/mounted-to) the
##          host on which this script runs, this script creates a list of 
##          files whose type contains a specified keyword, like 'executable'
##          or 'text' or 'data'.
##
##          The list has the 'basename' of the selected files separated into
##          column 1 --- so that the list can be sorted by 'basename'.
##          
##          One example of use of this utility (and sort method) is QUICKLY
##          finding copies of the same-named or similarly-named executables
##          in scattered sub-directories of a given directory.
##
##      See the 'Call Format' section below.
##############################################################################
## MAIN/ORIGINAL INTENT:
##           The main intent is to list duplicate executables  
##           in user-department/project directories (like /data/e46)
##           on a 'File Server' like 'engprd00'.  
##
##############################################################################
##
## CALLED BY:  'find_files4filetype_bygui'
##                             in $FEDIR/scripts
##             which is called by an option in the SpaceTools toolchest
##             invoked by 'spacetools' in $FEDIR/scripts.
##
##             May be called by nnsFEAmenu option 'u f ??', i.e.
##             feamain -> feautils.menu -> fileman.menu
##
##             Also can be called manually, for example ---
##             via login to 'engprd00'.
##                
##             Could be put in a 'ideasadm' cron job (in /apps/ideas/cron)
##             or in a 'caeadm' cron job (in /usr/people/caeadm/cron).
##
##############################################################################
## CALL FORMAT (examples):             See 'find_files4filetype_bygui'.
##                    
##      $FEDIR/scripts/set_localoutlist
##      $FEDIR/scripts/find_files4filetype  /data/e46 executable > $OUTLIST
##      xpg  $OUTLIST            
##                    
####  where the six positional arguments are used to set vars:
####                     TOPdir  fileTYPE
##                    
## OR ('rsh' form)
##      $FEDIR/scripts/set_localoutlist
##      rsh engprd00 $FEDIR/scripts/find_files4filetype \
##                                       /data/e46 executable > $OUTLIST
##      xpg  $OUTLIST            
##                
##############################################################################
##  POSSIBLE ENHANCEMENTS FOR MORE AUTOMATION:
##                
## 1a) With some userid selecting logic (and mail-address-finding/checking
##     logic), this script could automatically generate e-mail
##     to users who have many duplicate files in scattered sub-directories
##     under a given directory.
##
## 1b) Or the e-mail could be sent to 'ideasadm' or 'caeadm' or 'root' userids,
##     or specific Administrator ids (like 'bmo01' or 'tkj00')
##     for manual notification of the appropriate users.  The notification
##     of 'userids' could be done with a userid/to/mail-address lookup aid,
##     like a lookup aid in the 'mailtools' utility-cmd in $FEDIR/scripts.
##
###########################################################################
## NOTE: You can see only the non-commented, executable lines of this
##       script by using
##              egrep -v '^ *##|^ *# ' <script-name>
##       or 
##              grep -v '^ *##' <script-name> | grep -v '^ *# ' 
##############################################################################
## MAINTENANCE HISTORY:
## Written by: B.Montandon  O06 24Jul2000 Based on 'find_big_or_old_files4dir'
##                                        in   $FEDIR/scripts
##
## Updated by: Blaise Montandon 08apr2004 To avoid a bug in new SGI 'sort'
##                                        in IRIX 6.5.22, changed several cases
##                                        of  'sort -k5nr' to 'sort +4 -5nr'
##                                        and 'sort -k3nr' to 'sort +2 -3nr'.
##############################################################################

if test "$FEDIR" = ""
then
   FEDIR="/apps/nns_com/fea"
fi

HOST_ID=`hostname`

##############################################################################
## SET PARAMETERS TO BE USED BELOW.
##############################################################################
## NOTE: Defaults could be set for $1, $2 so that the script
##       can be tested quickly with no input parms.
##############################################################################

## FOR TESTING:
#  set -x

TOPdir="$1"
if test "$TOPdir" = ""
then
#  TOPdir="/data/e46"
#  TOPdir="/data/e46/Lessard"
   TOPdir="/data/e46/minerw/alt_ins3d"
#  exit
fi

########################################################################
## If the directory $DIRNAME is NSF-mounted (rather than local),
## show a message on possible missing data (due to local-parm on 'find'
## command) --- and probable slow performance --- i.e. do the
## query on a remote host-server.
########################################################################

NFSCHECK=`stat $TOPdir | grep nfs`
      
if test ! "$NFSCHECK" = ""
then

   CONFIRM_TEXT="\
Specified Directory: $TOPdir

is not local to this machine, $HOST_ID.

Try the query on a host/server for which the directory is local.
"

   export CONFIRM_TEXT

   CONFIRM_GEOM="+080+080"
   CONFIRM_MINSIZE="450 150"
   CONFIRM_TITLE="$WINTITLE"

   export CONFIRM_TITLE CONFIRM_MINSIZE CONFIRM_GEOM

   CONFIRM_COLOR="#3DADFF"
   export CONFIRM_COLOR

   CONFIRM_FONT="-adobe-helvetica-bold-r-normal--*-100-*-*-p-*-*-*"
   export CONFIRM_FONT

   JUNK=`$FEDIR/tkGUIs/confirm.tk DISMISS` &
   exit

fi


#############################################
## Figure out the num of dir-levels in $TOPdir.
#############################################

# TOPdirlevs=`echo "$TOPdir" | sed 's|/| |g' | wc -w`


#############################################
## SET THE 'fileTYPE' var.
#############################################

fileTYPE="$2"
if test "$fileTYPE" = ""
then
   fileTYPE="executable"
#  fileTYPE="text"
#  fileTYPE="data"
#  exit
fi

############################################################################
## NOTE ON 'ls -l' OUTPUT:
############################################################################
## In 'ls -l' output, if the file is NOT more than a year old,
## $1=perms $3=userid $5=bytes $6,$7=mmdd $8=hrs:min $NF=filename
############################################################################
## In 'ls -l' output, if the file IS more than a year old,
## $1=perms $3=userid $5=bytes $6,$7=mmdd $8=year $NF=filename
############################################################################


############################################################################
# REFMT_LS_CMD="eval awk '{printf (\"%10s %8s %12s %3s %2s %5s %s\n\", \$1, \$3, \$5, \$6, \$7, \$8, \$NF)}'"
## WORKS, but blanks in quoted format are compressed:
## eval awk '{printf ("%10s %8s %12s %3s %2s %5s %s
## ", $1, $3, $5, $6, $7, $8, $NF)}'
############################################################################
##  REFMT_LS_CMD="awk '{printf (\"%10s   %8s     %12s %3s %2s %5s %s\n\", \$1, \$3, \$5, \$6, \$7, \$8, \$NF)}'"
## DOES NOT WORK:
## awk '{printf ("%10s %8s %12s %3s %2s %5s %s
## ", $1, $3, $5, $6, $7, $8, $NF)}'
############################################################################
##  REFMT_LS_CMD='awk ''{printf ("%10s   %8s     %12s %3s %2s %5s %s\n", $1, $3, $5, $6, $7, $8, $NF)}'''
## DOES NOT WORK:
## awk {printf ("%10s %8s %12s %3s %2s %5s %s
## ", $1, $3, $5, $6, $7, $8, $NF)}
############################################################################

#  REFMT_LS_CMD="cut -c1-10,16-24,31-"

#  REFMT_LS_CMD="eval awk '{print \$1 \" \" \$3 \" \" \$5 \" \" \$6 \" \" \$7 \" \" \$8 \" \" \$NF}'"

#  REFMT_LS_CMD="eval awk '{printf (\"%-13s %-8s %12s %3s %2s %5s %s\n\", \$1, \$3, \$5, \$6, \$7, \$8, \$NF)}'"

# REFMT_LS_CMD="eval awk '{printf (\"%-13s %-8s %13.6f %3s %2s %5s %s\n\", \$1, \$3, \$5/1000000, \$6, \$7, \$8, \$NF)}'"

  REFMT_LS_CMD="eval awk '{printf (\"%-13s %-8s %13.6f %3s %2s %5s\n\", \$1, \$3, \$5/1000000, \$6, \$7, \$8)}'"

  LS_SORT_PREAWK="sort +4 -5nr"
# LS_SORT_PREAWK="sort -k5nr"

  LS_SORT_POSTAWK="sort +2 -3nr"
# LS_SORT_POSTAWK="sort -k3nr"

## FOR TESTING:
#    echo $REFMT_LS_CMD

#   REFMT_LS_HEADING="
#               Owner    File-Size    
# Permissions   Userid   GigMegKilByt Date-Time    Filename
# ----------    --------   |  |  |  | ------------ ------------------------------
# "

  REFMT_LS_HEADING="
                               Owner    FileSize(Meg)    
File Basename     Permissions  Userid   GigMeg.KilByt Date-Time    Directory
----------------  ----------   --------   |  |   |  | ------------ ------------------------------
"


#########################################################################
## SET REPORT FILENAME IN $OUTLIST.
#########################################################################

. $FEDIR/scripts/set_localoutlist
UNSORTEDLIST=${OUTLIST}_unsorted

#########################################################################
## PREP REPORT HEADING.
#########################################################################

   echo "
******************************** `date '+%Y %b %d  %a  %T%p %Z'` ******************
FILES WHICH ARE OF FILETYPE  ' $fileTYPE ',
IN THE VARIOUS SUB-DIRECTORIES
UNDER ${HOST_ID}:$TOPdir  :
                                                            (FILES BASENAME-SORTED)
$REFMT_LS_HEADING" > $OUTLIST


#########################################################################
## COLLECT FULLY-QUALIFIED FILENAMES FOR FILES WHICH MATCH $fileTYPE,
## AT ANY LEVEL UNDER THE  UNDER GIVEN $TOPdir:
#########################################################################

## FOR TESTING:
      set -x

# find $TOPdir  -local  -type f  -exec file {} \;  | grep "$fileTYPE"  >> $OUTLIST

FILELIST=`find $TOPdir  -local  -type f  -exec file {} \;  | grep "$fileTYPE" | \
                  cut -d: -f1`



#########################################################################
## COLLECT 'ls -l' INFO FOR THE COLLECTED FULLY-QUALIFIED FILENAMES.
#########################################################################

## FOR TESTING:
#     set -x

for FILE in $FILELIST
do
    BASENAME=`basename $FILE`
    DIRNAME=`dirname $FILE`
    FILEINFO=`ls -l $FILE | awk '{printf ("%-13s %-8s %13.6f %3s %2s %5s\n", $1, $3, $5/1000000, $6, $7, $8)}'"`

    #  echo "$BASENAME $DIRNAME `ls -l $FILE | $REFMT_LS_CMD`"

    BASENAME=`echo "$BASENAME                " | cut -c1-16`

    echo "$BASENAME $FILEINFO $DIRNAME" >> $UNSORTEDLIST

done

sort -k1 $UNSORTEDLIST >> $OUTLIST

## FOR TESTING:
#     echo "FindRetCode: $?"

## FOR TESTING:
#     set -


echo "

******************************** `date '+%Y %b %d  %a  %T%p %Z'` ******************
"  >> $OUTLIST


xpg $OUTLIST