Re^2: How to declare variables per loop

[Anonymous Monk]

Thank you very much!
I compose commands for gnuplot in the following way (it is all a single line):

plot $years $yscala "$abc_this_01" title "" with errorbars lt rgb "$co
+lor_abc" lw 2, "$abc_this_01" title "{/Helvetica=10 ABC_Stuff}" with 
+lines lt rgb "$color_abc" lw 3,"$def_this_01" title "" with errorbars
+ lt rgb "$color_def" lw 2, "$def_this_01" title "{/Helvetica=10 DEF_S
+tuff}" with lines lt rgb "$color_def" lw 3, "$ghi_this_01" title "" w
+ith errorbars lt rgb "$color_ghi" lw 2, "$ghi_this_01" title "{/Helve
+tica=10 GHI_Stuff}" with lines lt rgb "$color_ghi" lw 3, "$jkl_this_0
+1" title "" with errorbars lt rgb "$color_jkl" lw 2, "$jkl_this_01" t
+itle "{/Helvetica=10 JKL_Stuff}" with lines lt rgb "$color_jkl" lw 5,
+ "$mno_this_01" with steps lt 1 lw 5 title "{/Helvetica=10 Abnormalit
+y Border}"

and the files looks like this:

# Value Low High
2006 78.6 76.5 80.6
2007 86.3 84.7 87.6
2008 92.7 92.3 93.1
2009 95.3 94.9 95.6
2010 96.8 96.5 97.1
2011 97.0 96.5 97.5

It plots diagramms with the values (plus y-errorbars) from four sources (abc, def ...) and with the abnormality border. There are ca. 20 topics ("this", "that" etc.) with up to 8 subgroups (01, 02 etc.).
I have indeed the feeling that the composition is somehow ugly. Leastwise it works :-)
However I would be very glad for any suggestion how I could organize the data better.
Thanks again.
VE

[wfsp]

Blowing some white space into your command shows that there is a pattern that we might take advantage of.

plot $years $yscala 

"$abc_this_01" title "" with errorbars lt rgb "$color_abc" lw 2,
"$abc_this_01" title "{/Helvetica=10 ABC_Stuff}" with lines lt rgb "$c
+olor_abc" lw 3,

"$def_this_01" title "" with errorbars lt rgb "$color_def" lw 2,
"$def_this_01" title "{/Helvetica=10 DEF_Stuff}" with lines lt rgb "$c
+olor_def" lw 3,

"$ghi_this_01" title "" with errorbars lt rgb "$color_ghi" lw 2,
"$ghi_this_01" title "{/Helvetica=10 GHI_Stuff}" with lines lt rgb "$c
+olor_ghi" lw 3,

"$jkl_this_01" title "" with errorbars lt rgb "$color_jkl" lw 2,
"$jkl_this_01" title "{/Helvetica=10 JKL_Stuff}" with lines lt rgb "$c
+olor_jkl" lw 5,

"$mno_this_01" with steps lt 1 lw 5 title "{/Helvetica=10 Abnormality 
+Border}"

We could consider putting the data into a hash and then feed that into a template. I've used sprintf here but there are many alternatives. I've made some guesses about the actual spec.

#! /usr/perl/bin

use warnings;
use strict;

my %args = (
  years => 2011,
  yscala => q{xxx},
  sources => [
    {
      source => q{abc},
      color => q{red},
    },
    {
      source => q{def},
      color => q{blue},
    },
    {
      source => q{ghi},
      color => q{yellow},
    },
    {
      source => q{jkl},
      color => q{green},
    },
  ],
  topic => q{this},
  number => q{01},
  abnormality_border => q{mno},
);

my $cmd = build_command(%args);
print $cmd;

sub build_command {

  my %args = @_;
  
  my $cmd = sprintf(
    qq{plot %s %s\n\n}, 
    $args{years}, 
    $args{yscala}
  );
  
  for my $source (@{$args{sources}}){
  
    $cmd .= sprintf(
      qq{"C:/Stuff/%s_%s_%s.csv" title "" with errorbars lt rgb "%s_%s
+" lw 2,\n}, 
      $source->{source}, 
      $args{topic}, 
      $args{number}, 
      $source->{color},
      $source->{source},
    );

    $cmd .= sprintf(
      qq{"C:/Stuff/%s_%s_%s.csv" title "{/Helvetica=10 ABC_Stuff}" wit
+h lines lt rgb "%s_%s" lw 3,\n\n},
      $source->{source},
      $args{topic},
      $args{number},
      $source->{color},
      $source->{source},
    );
  }
  
  $cmd .= sprintf(
    qq{"C:/Stuff/%s_%s_%s.csv" with steps lt 1 lw 5 title "{/Helvetica
+=10 Abnormality Border}"},
    $args{abnormality_border},
    $args{topic},
    $args{number},
  );
  
  return $cmd;

}

output

plot 2011 xxx

"C:/Stuff/abc_this_01.csv" title "" with errorbars lt rgb "red_abc" lw
+ 2,
"C:/Stuff/abc_this_01.csv" title "{/Helvetica=10 ABC_Stuff}" with line
+s lt rgb "red_abc" lw 3,

"C:/Stuff/def_this_01.csv" title "" with errorbars lt rgb "blue_def" l
+w 2,
"C:/Stuff/def_this_01.csv" title "{/Helvetica=10 ABC_Stuff}" with line
+s lt rgb "blue_def" lw 3,

"C:/Stuff/ghi_this_01.csv" title "" with errorbars lt rgb "yellow_ghi"
+ lw 2,
"C:/Stuff/ghi_this_01.csv" title "{/Helvetica=10 ABC_Stuff}" with line
+s lt rgb "yellow_ghi" lw 3,

"C:/Stuff/jkl_this_01.csv" title "" with errorbars lt rgb "green_jkl" 
+lw 2,
"C:/Stuff/jkl_this_01.csv" title "{/Helvetica=10 ABC_Stuff}" with line
+s lt rgb "green_jkl" lw 3,

"C:/Stuff/mno_this_01.csv" with steps lt 1 lw 5 title "{/Helvetica=10 
+Abnormality Border}"

The initial hash (the data) could be kept in a separate file and read in by the script. I often find keeping the data and code separate can ease updating/maintainance. Taking it a bit further keeping the actual templates in a separate file wouldn't hurt either.

When you have it working nicely you would obviously take the white space out.

updated: added output and included the csv file.

[vagabonding electron]

Thank you wfsp, I put my answer at the end of the thread.
VE

[aartist]

You need Chart::Gnuplot. It should polish your interface. It is able to generate your chat on the fly. You can apply the loops here.