#!/bin/perl -w
#+
# datadirs -- print summary of recent data directories
#
# Purpose:
#	Locate and print the names of directories which contain FITS
#	images on the DEIMOS data disks, plus the number of blocks of 
#	data stored therein.
#
# Usage:
#	datadirs [-D] [-c] [account]
# 
# Arguments:
#	account = name of the account for which to search 
#		[default=current user, "all" gives all]
# 
# Switches:
#	-D = debug mode
#	-c = show compression status of directories.  Directories which 
#	     are not fully compressed are marked with an asterix
# 
# Output:
#	List of data directories
# 
# Restrictions:
#	None
# 
# Exit values:
#	 0 = normal completion
#	 1 = wrong number of arguments
#
# Example:
#	1) Print out list of all data directories:
#		polo{deimos4}85: datadirs all
#		    /dsk/iraid1/kics/2002aug17      290640 
#		    /dsk/iraid1/kics/2002aug18	   1853010 
#		      ...  
#		    /dsk/iraid1/deimos4/2002aug16 30691736 
#		    /dsk/iraid1/deimos4/2002aug15  2696594 
#		
# 	2) Print out list of all data directories for this account:
#		polo{deimos4}87: datadirs
#		    /dsk/iraid1/deimos4/2002aug16 30691736 
#		    /dsk/iraid1/deimos4/2002aug15  2696594
#
# 	3) Print out list of all occupied data directories for "deimos3":
#		polo{deimos4}87: datadirs deimos3
#		    /dsk/iraid1/deimos3/2002aug13 37433128 
#		    /dsk/iraid1/deimos3/2002aug14 24779988 
#		    /dsk/iraid2/deimos3/2002aug09  5903164 
#		    /dsk/iraid2/deimos3/2002aug10 24388340 
#		    /dsk/iraid2/deimos3/2002aug11 21018734 
#		    /dsk/iraid2/deimos3/2002aug12 35955228 
#		    /dsk/iraid2/deimos3/2002aug13       44 
#		    /dsk/iraid2/deimos3/2002aug14       44
# 
# 	4) Print out list of all data directories and show status
#	of compressed files:
#		[61] deimos@polo> datadirs -c all
#		/dsk/iraid2/deimos7/2002oct01       5158306  53/ 53 FITS images compressed
#		/dsk/iraid2/deimos7/2002oct02       4084054  39/ 39 FITS images compressed
#		/dsk/iraid2/deimos3/2002oct03       7462189  95/ 96 FITS images compressed *
#		/dsk/iraid2/deimos3/2002oct04      15981853  37/140 FITS images compressed *
#		/dsk/iraid2/deimos3/2002oct05       7764075   0/ 78 FITS images compressed *
#		/dsk/iraid1/deimos3/2002oct06       4966609  85/ 86 FITS images compressed *
#		/dsk/iraid2/deimos8/2002oct07       7716172  99/ 99 FITS images compressed
#		/dsk/iraid2/deimos8/2002oct08       6975709  96/ 96 FITS images compressed
#-
# Modification history:
#	2002-Aug-21	GDW	Original version
#	2002-Oct-10	GDW	Converted to Perl and added date sorting
#	2002-Oct-16	GDW	Added "-c" switch
#	2002-Oct-18	GDW	Changes to support /sdata
#	2003-May-22	GDW	Redirect cruft from "find" to /dev/null
#	2004-Dec-10	GDW	Change globs to readdirs to avoid errors
#				resulting from large directories
#	2005-Sep-12	GDW	Build disk list automatically
#	2012-May-25	GDW	Remove kics from list and redo find
#	2014-Jun-22	GDW	Change pattern match from 2?????? to 20*
#-----------------------------------------------------------------------

use strict;
use Getopt::Std;

my( $opt_c, $opt_D);
my( $i);
my( $disk);
my( @accounts) = $ENV{USER};
my( @files, @gzfits, @fits);
my( @good_dirs);
my( @tail);
my( $n_fits, $n_gzfits);
my( $debug) = 0;
my( $account, $topdir);
my( @dirs);
my( $dir);
my( $size);

if ( $#ARGV >= 0 ){
  @accounts = @ARGV
}

# need this to avoid warnings about opt_c used only once...
getopts("cD");
if( $opt_D){$debug=1};

# check for "all" option...
if ( $accounts[0] eq "all" ){
  @accounts = qw(deimos dmoseng);
  for ( $i=1 ; $i<=20 ; $i++ ){
    push @accounts, "deimos$i";
  }
}

# build disk list...
$_ = `/usr/sbin/df`;
my( @buf) = split;
die "No output from df" if $#buf < 1;
my( @disks) = grep s/^(\/sdata....)/$1/, @buf;
die "No data disks found" if $#disks < 0;
@disks = sort @disks;

# loop over accounts...
foreach $account ( @accounts ){
  print "Checking account $account\n" if $debug;
  foreach $disk ( @disks ){
    print "  Checking disk $disk\n" if $debug;

    # find directories which belong to this account...
    $topdir = "$disk/$account";
    next if not -e $topdir;
    $_ = `/bin/find $topdir/20* -prune -type d 2>/dev/null`;
    @dirs = split;
    next if $#dirs < 0;
	    
    # eliminate directories which have no fits files...
    foreach $dir ( @dirs ){

      # skip dirs with "kroot"...
      next if $dir =~ m/kroot/;

      print "    Checking dir $dir\n" if $debug;
      opendir DATADIR, $dir;
      @files = grep /\.fits/, readdir DATADIR;
      closedir DATADIR;
      if ( $#files > 0 ){
	push @good_dirs, $dir;
	$dir =~ m|^.*/(.*)$|;
	push @tail, $1;
      }
    }
  }
}

# sort directories by date...
my( @keys) = (0..@good_dirs-1);
my( @sortkeys) = sort { ymd2num($tail[$a]) <=> ymd2num($tail[$b]) } @keys;

# print results...
foreach $i (@sortkeys){
  $dir = $good_dirs[$i];
  $_ = `du -sk $dir`;
  ($size, $dir) = split;
  printf "%-30s %12s", $dir, $size;
  if ( $opt_c ){

    # read list of all FITS files...
    opendir DATADIR, $dir;
    @files = grep /\.fits/, readdir DATADIR;
    closedir DATADIR;

    # count gzipped-FITS files...
    @gzfits = grep /\.fits.*gz$/, @files;
    $n_gzfits = $#gzfits + 1;

    # count FITS files...
    @fits = grep /\.fits$/, @files;
    $n_fits = $n_gzfits + $#fits + 1;

    printf " %3d/%3d FITS images compressed", $n_gzfits, $n_fits;
    if ( $n_gzfits != $n_fits ) { printf " *" }
  }
  printf "\n";
}
exit;

#-----------------------------------------------------------------------
sub ymd2num{
#-----------------------------------------------------------------------
# Purpose:
#	Convert a string of form yyyyMMMdd to a numeric value that
#	can be sorted.
#-----------------------------------------------------------------------
    use Time::Local;
    my( $buf) = @_;
    my( $seconds, $minutes, $hours) = (0,0,0);
    my( $year, $month, $day);
    my( $foo);

    my( %monthval);
    $monthval{'jan'} = 1;
    $monthval{'feb'} = 2;
    $monthval{'mar'} = 3;
    $monthval{'apr'} = 4;
    $monthval{'may'} = 5;
    $monthval{'jun'} = 6;
    $monthval{'jul'} = 7;
    $monthval{'aug'} = 8;
    $monthval{'sep'} = 9;
    $monthval{'oct'} = 10;
    $monthval{'nov'} = 11;
    $monthval{'dec'} = 12;
    
    if ( $buf =~ /(\d\d\d\d)(\w\w\w)(\d\d)/ ) {
	$year = $1;
	$month = $2;
	$day = $3;
	if( defined($monthval{$month})){
	  return timegm( $seconds, $minutes, $hours, $day, 
			 $monthval{$month}-1, $year-1900)
	} else {
	  return 0
	}
      } else {
	return 0
      }
  }