NAME

    Graphics::Skullplot - Plot the result of an SQL select (e.g. from an
    emacs shell window)

VERSION

    Version 0.01

SYNOPSIS

       # To use this from emacs, see scripts/skullplot.el.
       # That elisp code accesses the perl script: scripts/skullplot.pl
    
       # the code used by skullplot.pl
       my $plot_hints = { indie_count           => $indie_count,
                          dependent_requested   => $dependent_requested,
                          independent_requested => $independent_requested,
                        };
       my %gsp_args = 
         ( input_file   => $dbox_file,
           plot_hints   => $plot_hints, );
       $gsp_args{ working_area } = $working_area if $working_area;
       $gsp_args{ image_viewer } = $image_viewer if $image_viewer;
       my $gsp = Graphics::Skullplot->new( %gsp_args );
    
       $gsp->show_plot_and_exit();  # does an exec 

DESCRIPTION

    Graphics::Skullplot is a module that works with the result from a
    database select in the common tabular text "data box" format. It has
    routines to generate and display plots of the data in png format.

    Internally it uses the Table::BoxFormat module to parse the text table,
    and the Graphics::Skullplot::ClassifyColumns module to determine the
    types of the columns.

    The default image viewer is the ImageMagick "display" command.

    The immediate use for this code is to act as the back-end for the
    included Emacs package scripts/skullplot.el, so that database select
    results generated in an emacs shell window can be immediately plotted.

    This elisp code calls scripts/skullplot.pl, which might be used in
    other contexts.

METHODS

    new

      Creates a new Graphics::Skullplot object. Object attributes:

      working_area

	Scratch location where intermediate files are created. Defaults to
	"/tmp".

      image_viewer

	Defaults to 'display', the ImageMagick viewer (a dependency on
	Image::Magick ensures it's available)

    builder methods (largely for internal use)

      builder_image_viewer Currently just returns a hardcoded selection
      (the ImageMagick "display" program).

    generate_output_filenames

      Example usage:

        # relies on object settings: "input_file" and "working area"
        my $fn = 
          generate_filenames();
        my $basename = $fn->{ base };
        # full paths to file in $working_area
        my $tsv_file  = $fn->{ tsv };  
        my $png_file  = $fn->{ png };  

    plot_tsv_to_png

      Generate the r-code to plot the tsv file data as the png file. Takes
      one argument, a hash of "field metadata".

      The file names (tsv, png, plus internal formats) come from the
      "naming" object field.

      Example usages:

        $self->plot_tsv_to_png( $plot_cols ); 

    generate_png_file

      Example usage:

        $self->generate_png_file( $pc, $fn );

      Runs the given plot code (first argument) using the file-name
      metadata (second argument, defaults to object's naming), saving the
      plot as a png file ($fn->{png}).

      This generates a file of R code to run with an Rscript call. In debug
      mode, this generates a standalone unix script. ($DEBUG).

    display_png_and_exit

      Open the given png file in an image viewer Defaults to "png" field in
      object's "naming".

      This internally does an exec: it should be the last thing called.

      The image viewer can be set as the second, optional field. The
      default image viewer is ImageMagick's "display".

      Example usage:

        my $naming = $self->naming;
        my $png_file = $naming->{ png };
        $self->display_png_and_exit( $png_file );

    show_plot_and_exit

      The method called by the skullplot.pl script to actually plot the
      data from a "data box format" file, using the plot_hints.

      It's expected that the dbox file (input_file) and the plot_hints will
      be defined at object creation, but at present those settings may be
      overridden here and given as first and second arguments.

      This should be used at the end of the program (internally it does an
      "exec").

    classify_columns

      Given a reference to the tabular data in the form of an array of
      arrays, returns metadata for each column to be used in deciding how
      to plot the data.

      Example usage:

        my $plot_cols = $self->classify_columns( $data );

      Classify the columns from the tabular data, returning a
      "fields_metadata" hash ref.

      This is a wrapper around a provisional technique to make it easier to
      swap in better ones later.

      At present, the metadata fields are:

           x           => $x_field  (( rename indie_x ))
           y           => $y_field
           gb_cats      => [ @gb_cats ]
           dependents_y => [ @dependents_y ]

    dumporama

      Report on state of object fields.

    fryhash

NOTES

 TODO

      * Limited to two group by categories (in addition to the x-axis):
      used with colour & shape If there's more than 2, fuse them together
      into a compound, use with colour

      * See R Graphics Cookbook, p.205: setting up the tics and labels.

          $pc .= 'p + scale_x_date';
          $pc .= '';

      * Currently this defaults to viewing images using the "display"
      program. Alternately, the builder_image_viewer could scan through a
      list of likely viewers and pick the first that's installed.

AUTHOR

    Joseph Brenner, <doom@kzsu.stanford.edu>, 16 Nov 2016

COPYRIGHT AND LICENSE

    Copyright (C) 2016 by Joseph Brenner

    This program is free software; you can redistribute it and/or modify it
    under the terms of either: the GNU General Public License as published
    by the Free Software Foundation; or the Artistic License.

    See http://dev.perl.org/licenses/ for more information.