bin/data_stats

   1 #!/usr/bin/perl
   2
   3 use FindBin qw($Bin);
   4 use lib "$Bin/../lib";
   5
   6 # Don't edit the line below, it must look exactly like this.
   7 # Everything above this line will be replaced #
   8
   9 use PsN;
  10 use data;
  11 use debug;
  12 use strict;
  13 use Getopt::Long;
  14
  15 my ($opt_help,
  16     $opt_long_help,
  17     $opt_columns,
  18     $opt_ignore_columns,
  19     $opt_debug,
  20     $opt_debug_package,
  21     $opt_debug_subroutine);
  22
  23 my $opts = { "h|?"                => \$opt_help,
  24              "help"               => \$opt_long_help,
  25              "columns:s"          => \$opt_columns,
  26              "ignore_columns:s"   => \$opt_ignore_columns,
  27              "debug:0"            => \$opt_debug,
  28              "debug_package:s"    => \$opt_debug_package,
  29              "debug_subroutine:s" => \$opt_debug_subroutine };
  30
  31 my $res = GetOptions( %{$opts} );
  32
  33 exit unless $res;
  34
  35 if ( scalar( @ARGV ) < 1 and !($opt_help or $opt_long_help) ){
  36   print "At least one list file must be specified. Use 'data_stats -h' for help.\n";
  37   exit;
  38 }
  39
  40
  41 if($opt_help or $opt_long_help) {
  42   print <<'ENDHELP';
  43
  44   data_stats
  45
  46   Usage:
  47
  48   data_stats [ -h | -? ] [ -help ]
  49              [ -columns='string' ]
  50              [ -ignore_columns='string' ]
  51              [ -debug='integer' ]
  52              [ -debug_package='string' ]
  53              [ -debug_subroutine='string' ]
  54              outputfile(s)
  55
  56 ENDHELP
  57
  58   if( $opt_long_help and !$opt_help ){
  59
  60     print <<'ENDHELP';
  61   Description:
  62
  63     data_stats calculates and prints simple statistics for data
  64     files. Simple stastics are max, min, mean, median, range and
  65     standard deviation for each column.
  66
  67   Examples:
  68
  69     Calculate statistics for all columns in file.dta
  70
  71     $ data_stats file.dta
  72
  73     Calculate statistics for all columns except WGT and AGE.
  74
  75     $ data_stats --ignore_columns=WGT,AGE
  76
  77     Calculate statistics for WGT and AGE only
  78
  79     $ data_stats --columns=WGT,AGE
  80
  81   Options:
  82
  83     The options are given here in their long form. Any option may be
  84     abbreviated to any nonconflicting prefix. The -threads option may
  85     be abbreviated to -t(or even -thr) but -debug may not be
  86     abbreviated to -d because it conflicts with -debug_packages and
  87     -debug_subroutines.
  88
  89     The following options are valid:
  90
  91     -h | -?
  92
  93     Print a list of options.
  94
  95
  96     -columns='strings'
  97
  98     By default data_stats prints statistics for all columns of the
  99     data set. But if you give a comma separated list of column headers
 100     or column numbers with the -columns option, then data_stats will
 101     print statistics for those columns only.
 102
 103
 104     -ignore_columns='strings'
 105
 106     ignore_columns work the oposite way of -columns, it lets you
 107     select column headers or column numbers for which data_stats
 108     should not print statistcs. It also takes a comma separated list.
 109
 110     -help
 111
 112     Print this, longer, help message.
 113
 114
 115     -debug='integer'
 116
 117     Default value is: 0
 118
 119     This is mainly for developers who whish to debug PsN. By default
 120     'integer' is zero but you can try setting it to '1' and you might
 121     get some helpfull warnings. If you run in to problems that require
 122     support, you might be told to crank this number up and send the
 123     output to us.
 124
 125
 126     -debug_package='string'
 127
 128     Default value is: empty string
 129
 130     If use together with '-debug' it is possible to choose which part
 131     of PsN you want to see debug messages from. Again this is mostly
 132     for developers.
 133
 134
 135     -debug_subroutine='string'
 136
 137     Default value is: empty string
 138
 139     With this option it is possible to specify, with even finer
 140     granularity, which part of PsN you want to see debug messages
 141     from. This is definitly only for developers.
 142
 143 ENDHELP
 144   }
 145   exit;
 146 }
 147
 148 debug -> level( $opt_debug );
 149 debug -> package( $opt_debug_package );
 150 debug -> subroutine( $opt_debug_subroutine );
 151
 152 my $datafile = $ARGV[0];
 153
 154 my $dataobj = data -> new ('filename'=> $datafile);
 155
 156 my %ignore_columns;
 157
 158 if( $opt_ignore_columns ){
 159   foreach my $ign ( split( /,/ , $opt_ignore_columns ) ){
 160     $ignore_columns{$ign} = 1;
 161   }
 162 }
 163
 164 my @columns;
 165
 166 if( $opt_columns ) {
 167   @columns = split( /,/, $opt_columns)
 168 } else {
 169   @columns = @{$dataobj -> header};
 170 }
 171
 172 my $first = 1;
 173 foreach my $function ( ' 'x8, 'Min', 'Max', 'Mean', 'Median', 'Range', 'SD', ){
 174   printf "%-8s", $function;
 175   foreach my $head ( @columns ){
 176     unless( $ignore_columns{$head} ){
 177       if( $first ){
 178         printf "%-8s", $head;
 179       } else {
 180         my $function = lc( $function );
 181         if( $head =~ /[^\d]/ ){
 182           printf "%-8d", $dataobj -> $function( column_head => $head );
 183         } else {
 184           printf "%-8d", $dataobj -> $function( column => $head );
 185         }
 186       }
 187     }
 188   }
 189   $first = 0;
 190   print "\n";
 191 }
 192
 193
 194 print "\n";