Draft of 'Getting Started With Matplotlib's OO Class Library'

I took up John's suggestion to 'new users starting on the path to matplotlib OO API enlightenment to make notes and write a tutorial as you go'.

I'm posting it to the list to generate some feedback and hopefully help out any fellow newbies (although I would prefer it to be on a Wiki somewhere).

Feel free to comment....

Robert

···

==================================================
Getting Started With Matplotlib's OO Class Library

Introduction
------------

    For those people coming to Matplotlib without any prior experience
    of MatLab and who are familiar with the basic concepts of
    programming API's and classes, learning to use Matplotlib via the
    class library is an excellent choice. The library is well put
    together and works very intuitively once a few fundamentals are
    grasped.

    The advice from John Hunter, the original developer of the library
    is 'don't be afraid to open up matplotlib/pylab.py to see how the
    pylab interface forwards its calls to the OO layer.' That in
    combination with the user's guide, the examples/embedding demos,
    and the mailing lists, which are regularly read by many developers
    are an excellent way to learn the class library.

    Following is a brief introduction to using the class library,
    developed as I came to grips with how to produce my first graphs.

Classes/Terms
-------------

    FigureCanvas - is primarily a container class to hold the Figure
    instance, an approach which enforces a rigid segregation between
    the plot elements, and the drawing of those elements. It can be
    loosely thought of as 'the place where the ink goes'.

    Figure - a container for one or more Axes. It is possible to
    create and manage an arbitrary number of figures using the Figure
    class. Note also that a figure can have its own line, text and
    patch elements, independent of any axes.

    Axes - the rectangular area which holds the basic elements (Line2D,
    Rectangle, Text, etc) that are generated by the Axes plotting
    commands (e.g. the Axes plot, scatter, bar, hist methods). The Figure
    methods add_axes and add_subplot are used to create and add an Axes
    instance to the Figure. You should not instantiate an Axes instance
    yourself, but rather use one of these helper methods.

    Line - implemented in the Line2D class, can draw lines(!) with a
    variety of styles (solid, dashed, dotted, etc), markers (location
    indicators on the line - point, circle, triangle, diamond, etc) and
    colours (k or #000000 - black, w or #FFFFFF - white, b or #000099 -
    blue, etc)

    Text - a class that handles storing and drawing of text in window
    or data coordinates. The text can be coloured, rotated, aligned in
    various ways relative to the origin point, and have font properties
    (weight, style, etc) assigned to it.

    Patch - a patch is a two dimensional shape with a separately
    specifiable face and edge colour. Specialised patch classes include
    circle, rectangle, regular polygon and more.

    Ticks - the indicators of where on an axis a particular value
    lies. Separate classes exist for the x and y axis ticks, (XTick,
    YTick) and each are comprised of the primitive Line2D and Text
    instances that make up the tick.

    Artist - Everything that draws into a canvas derives from Artist
    (Figure, Axes, Axis, Line2D, Rectangle, Text, and more). Some of
    these are primitives (Line2D, Rectangle, Text, Circle, etc) in that
    they do not contain any other Artists, some are simple composites,
    e.g. XTick which is mad up of a couple of Line2D and Text instances
    (upper and lower tick lines and labels), and some are complex, e.g.
    and Axes or a Figure, which contain both composite and primitive
    artists.

Techniques
----------

1. Setting up an agg backend canvas:

    from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
    from matplotlib.figure import Figure

    fig = Figure()
    canvas = FigureCanvas(fig)

2. To set the size of the Figure, use the figsize keyword, which uses
    inches as the units:

    # this results in a 204x192 pixel png - if output at 100 dpi, using
    # canvas.print_figure(.., dpi=100)
    fig = Figure(figsize=(2.04,1.92))
    canvas = FigureCanvas(fig)

3. To add a single subplot:

    # The 111 specifies 1 row, 1 column on subplot #1
    ax = fig.add_subplot(111)

4. To change the axes size and location on construction, e.g to fit
    the labels in on a small graph:

    ax = fig.add_axes([0.2,0.2,0.5,0.7])

    An existing Axes position/location can be changed by calling
    the set_position() method.

5. Adding a graph of some sort is as simple as calling the required
    function on the axes instance:

    p1 = ax.bar(...) or p1 = ax.plot(...)

6. Setting a label with extra small text:

    ax.set_xlabel('Yrs', size='x-small')

7. Setting the graph title:

    ax.set_title(A graph title', size='x-small')

8. To enable only the horizontal grid on the major ticks:

    ax.grid(False)
    ax.yaxis.grid(True, which='major')

9. To only have a left y-axis and a bottom x-axis:

    # set the edgecolor and facecolor of the axes rectangle to be the same
    frame = ax.axesPatch
    frame.set_edgecolor(frame.get_facecolor())
    # Specify a line in axes coords to represent the left and bottom axes.
    bottom = Line2D([0, 1], [0, 0], transform=ax.transAxes)
    left = Line2D([0, 0], [0, 1], transform=ax.transAxes)
    ax.add_line(bottom)
    ax.add_line(left)

10. To change the size of the tick labels :

     labels = ax.get_xticklabels() + ax.get_yticklabels()
     for label in labels:
         label.set_size('x-small')

11. Removing the black rectangle around an individual bar graph
     rectangle (by changing it to the bar colour) :

     c = '#7FBFFF'
     p1 = ax.bar(ind, values, width, color=c)
     for r in p1:
         r.set_edgecolor(c)

Putting it together
-------------------

     Following is a simple example of how to use the class library
     This is examples/agg_oo.py in the matplotlib src distribution, also
     found (like all examples) at http://matplotlib.sf.net/examples

     #!/usr/bin/env python
     """
     A pure OO (look Ma, no pylab!) example using the agg backend
     """
     from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
     from matplotlib.figure import Figure

     fig = Figure()
     canvas = FigureCanvas(fig)
     ax = fig.add_subplot(111)
     ax.plot([1,2,3])
     ax.set_title('hi mom')
     ax.grid(True)
     ax.set_xlabel('time')
     ax.set_ylabel('volts')
     canvas.print_figure('test')