SVG::TT::Graph::TimeSeries - Create presentation quality SVG line graphs of time series easily

  use SVG::TT::Graph::TimeSeries;

  my @data_cpu = ('2003-09-03 09:30:00',23,'2003-09-03 09:45:00',54,'2003-09-03 10:00:00',67,'2003-09-03 10:15:00',12);
  my @data_disk = ('2003-09-03 09:00:00',12,'2003-09-03 10:00:00',26,'2003-09-03 11:00:00',23);

  my $graph = SVG::TT::Graph::TimeSeries->new({
        'height' => '500',
        'width' => '300',
  });

  $graph->add_data({
    'data' => \@data_cpu,
        'title' => 'CPU',
  });

  $graph->add_data({
    'data' => \@data_disk,
        'title' => 'Disk',
  });

  print "Content-type: image/svg+xml\r\n\r\n";
  print $graph->burn();

This object aims to allow you to easily create high quality SVG line graphs of time series. You can either use the default style sheet or supply your own. Either way there are many options which can be configured to give you control over how the graph is generated - with or without a key, data elements at each point, title, subtitle etc. All times must be given a format parseable by HTTP::Date. The DateTime module is used for all date/time calculations.

Note that the module is currently limited to the Unix-style epoch-based date range limited by 32 bit signed integers (around 1902 to 2038).

  use SVG::TT::Graph::TimeSeries;

  my $graph = SVG::TT::Graph::TimeSeries->new({

    # Optional - defaults shown
    'height'            => 500,
    'width'             => 300,

    'show_y_labels'     => 1,
    'scale_divisions'   => '',
    'min_scale_value'   => 0,
    'max_scale_value'   => '',

    'show_x_labels'     => 1,    
    'timescale_divisions' => '',
    'min_timescale_value' => '',
    'max_timescale_value' => '',
    'x_label_format'    => '%Y-%m-%d %H:%M:%S',
    'stagger_x_labels'  => 0,
    'rotate_x_labels'   => 0,

    'show_data_points'  => 1,
    'show_data_values'  => 1,
    'rollover_values'   => 0,

    'area_fill'         => 0,

    'show_x_title'      => 0,
    'x_title'           => 'X Field names',

    'show_y_title'      => 0,
    'y_title'           => 'Y Scale',

    'show_graph_title'      => 0,
    'graph_title'           => 'Graph Title',
    'show_graph_subtitle'   => 0,
    'graph_subtitle'        => 'Graph Sub Title',
    'key'                   => 0,
    'key_position'          => 'right',

    # Optional - defaults to using internal stylesheet
    'style_sheet'       => '/includes/graph.css',
  });

The constructor takes a hash reference with values defaulted to those shown above - with the exception of style_sheet which defaults to using the internal style sheet.

  my @data_cpu = ('2003-09-03 09:30:00',23,'2003-09-03 09:45:00',54,'2003-09-03 10:00:00',67,'2003-09-03 10:15:00',12);
  or
  my @data_cpu = (['2003-09-03 09:30:00',23],['2003-09-03 09:45:00',54],['2003-09-03 10:00:00',67],['2003-09-03 10:15:00',12]);

  $graph->add_data({
    'data' => \@data_cpu,
    'title' => 'CPU',
  });

This method allows you to add data to the graph object. The data is expected to be a list of time, value pairs. It can be called several times to add more data sets in.

  my $graph->clear_data();

This method removes all data from the object so that you can reuse it to create a new graph but with the same config options.

  print $graph->burn();

This method processes the template with the data and config which has been set and returns the resulting SVG.

This method will croak unless at least one data set has been added to the graph object.

  my $value = $graph->method();
  my $confirmed_new_value = $graph->method($value);

The following is a list of the methods which are available to change the config of the graph object after it has been created.

height()

Set the height of the graph box, this is the total height of the SVG box created - not the graph it self which auto scales to fix the space.

width()

Set the width of the graph box, this is the total width of the SVG box created - not the graph it self which auto scales to fix the space.

style_sheet()

Set the path to an external stylesheet, set to '' if you want to revert back to using the defaut internal version. To create an external stylesheet create a graph using the default internal version and copy the stylesheet section to an external file and edit from there.

show_data_values()

Show the value of each element of data on the graph.

rollover_values()

Shows data values and data points when the mouse is over the point. Used in combination with show_data_values and/or show_data_points.

data_value_format()

Format specifier to for data values (as per printf).

show_data_points()

Show a small circle on the graph where the line goes from one point to the next.

max_time_span()

Maximum timespan for a line between data points. If this span is exceeded, the points are not connected. This is useful for skipping missing data sections. The expected form is: '<integer> [years | months | days | hours | minutes | seconds]'

stacked()

Accumulates each data set. (i.e. Each point increased by sum of all previous series at same time). Default is 0, set to '1' to show. All data series have the same number of points and must have the same sequence of time values for this option.

min_scale_value()

The point at which the Y axis starts, defaults to '0', if set to '' it will default to the minimum data value.

max_scale_value()

The point at which the Y axis ends, if set to '' it will default to the maximum data value.

scale_divisions()

This defines the gap between markers on the Y axis, default is a 10th of the range, e.g. you will have 10 markers on the Y axis. NOTE: do not set this too low - you are limited to 999 markers, after that the graph won't generate.

show_x_labels()

Whether to show labels on the X axis or not, defaults to 1, set to '0' if you want to turn them off.

x_label_format()

Format string for presenting the X axis labels. The POSIX strftime() function is used for formatting (see strftime man pages and LC_TIME locale information).

show_y_labels()

Whether to show labels on the Y axis or not, defaults to 1, set to '0' if you want to turn them off.

y_label_format()

Format string for presenting the Y axis labels (as per printf).

timescale_divisions()

This defines the gap between markers on the X axis. Default is the entire range (only start and end axis labels). The expected form is: '<integer> [years | months | days | hours | minutes | seconds]' The default time period if not provided is 'days'. These time periods are used by the DateTime::Duration methods.

stagger_x_labels()

This puts the labels at alternative levels so if they are long field names they will not overlap so easily. Default it '0', to turn on set to '1'.

rotate_x_labels()

This turns the X axis labels by 90 degrees. Default it '0', to turn on set to '1'.

min_timescale_value()

This sets the minimum timescale value (X axis). Any data points before this time will not be shown. The date/time is expected in ISO format: YYYY-MM-DD hh:mm:ss.

max_timescale_value()

This sets the maximum timescale value (X axis). Any data points after this time will not be shown. The date/time is expected in ISO format: YYYY-MM-DD hh:mm:ss.

show_x_title()

Whether to show the title under the X axis labels, default is 0, set to '1' to show.

x_title()

What the title under X axis should be, e.g. 'Months'.

show_y_title()

Whether to show the title under the Y axis labels, default is 0, set to '1' to show.

y_title()

What the title under Y axis should be, e.g. 'Sales in thousands'.

show_graph_title()

Whether to show a title on the graph, default is 0, set to '1' to show.

graph_title()

What the title on the graph should be.

show_graph_subtitle()

Whether to show a subtitle on the graph, default is 0, set to '1' to show.

graph_subtitle()

What the subtitle on the graph should be.

key()

Whether to show a key, defaults to 0, set to '1' if you want to show it.

key_position()

Where the key should be positioned, defaults to 'right', set to 'bottom' if you want to move it.

The default stylesheet handles upto 12 data sets, if you use more you must create your own stylesheet and add the additional settings for the extra data sets. You will know if you go over 12 data sets as they will have no style and be in black.

For examples look at the project home page http://leo.cuckoo.org/projects/SVG-TT-Graph/

None by default.

Leo Lapworth for SVG-TT-Graph. Stephen Morgan for creating the TT template and SVG.

David Meibusch (David.Meibusch@adc.com)

SVG::TT::Graph, SVG::TT::Graph::Line, SVG::TT::Graph::Bar, SVG::TT::Graph::BarHorizontal, SVG::TT::Graph::BarLine, SVG::TT::Graph::Pie