Plot Container

This examples shows how you would currently generate a plot that shows projections of a 2D image.

While it does work, this is for development purposes – while the figure list currently works well for collecting a series of individual matplotlib plots along with text and generating a notebook presentation from them, things that involve multiple matplotlib Axes, like this one, are unnecessarily complicated.

The comments describe how the new proposed objects will work.

Traceback (most recent call last):
  File "C:\Users\johnf\notebook\pyspecdata\examples\plot_container.py", line 106, in <module>
    adjust_spines(axTop, ["left"])
  File "C:\Users\johnf\notebook\pyspecdata\examples\plot_container.py", line 26, in adjust_spines
    spine.set_smart_bounds(True)
AttributeError: 'Spine' object has no attribute 'set_smart_bounds'

from numpy import *
from pyspecdata import *
import matplotlib.pyplot as plt
from matplotlib.ticker import NullFormatter
from matplotlib import transforms

# {{{ from spine_placement_demo
def adjust_spines(ax, spines):
    for loc, spine in ax.spines.items():
        if loc in spines:
            spine.set_position(("outward", 10))  # outward by 10 points
            spine.set_smart_bounds(True)
        else:
            spine.set_color("none")  # don't draw spine

    # turn off ticks where there is no spine
    if "left" in spines:
        ax.yaxis.set_ticks_position("left")
    else:
        # no yaxis ticks
        ax.yaxis.set_ticks([])

    if "bottom" in spines:
        ax.xaxis.set_ticks_position("bottom")
    if "top" in spines:
        ax.xaxis.set_ticks_position("top")
    else:
        # no xaxis ticks
        ax.xaxis.set_ticks([])


# }}}

# based off an example from matplotlib gallery

# the random data
test_data = nddata(
    random.normal(size=100 * 100) + 1j * random.normal(size=100 * 100),
    [100, 100],
    ["x", "y"],
)
test_data.setaxis("x", "#").setaxis("y", "#")
test_data.set_units("x", "s").set_units("y", "m")
test_data.reorder("y")

# definitions for the axes
left, width = 0.12, 0.65
bottom, height = 0.15, 0.65
padding = 0.05
bottom_side = bottom + height + padding
left_side = left + width + padding

rect_scatter = [left, bottom, width, height]
rect_top = [left, bottom_side, width, 1.0 - bottom_side - padding]
rect_right = [left_side, bottom, 1.0 - left_side - padding, height]

with figlist_var() as fl:
    # A lot of extra junk here b/c figure list not set up for multiple matplotlib axes.
    #
    # The new idea is to create a container, which in this case would be all
    # three of these  axes together.
    #
    # When we drop data into the container, it would automatically decide what
    # to do with it, based on its shape (here, 2D data would go in the center,
    # while 1D data would go to one of the outside plots), and would directly
    # use matplotlib commands
    # -- so, there would be no need to pass the figure to "next", set the
    # various titles to 0, etc.
    #
    # also, the container would just hold a list of data until its actually
    # read to render the plots -- it would *then* do the human units thing, so
    # that there were no human units error
    #
    # the domain overview plot would be 1 matplotlib axes object for each
    # coherence pathway and would probably take advantage of this command --
    # https://matplotlib.org/examples/axes_grid/simple_axesgrid2.html

    # start with a rectangular Figure
    fig = plt.figure(1, figsize=(9, 5.56))
    axCentral = plt.axes(rect_scatter)
    axRight = plt.axes(rect_right)
    axTop = plt.axes(rect_top)
    fl.next("test figure", fig=fig, ax=axCentral)

    fl.image(test_data, ax=axCentral, human_units=False)
    axCentral.set_aspect('auto')
    axCentral.set_title("")
    fl.plot(test_data.C.sum("y"), ax=axTop, human_units=False)
    axTop.autoscale(enable=True, tight=True)  # axis tight
    axTop.set_ylabel("")
    axTop.set_xlabel("")
    adjust_spines(axTop, ["left"])
    base = axRight.transData
    rot = transforms.Affine2D().rotate_deg(90)
    fl.plot(test_data.C.sum("x"), transform=(rot + base), ax=axRight, human_units=False)
    axRight.set_ylabel("")
    axRight.set_xlabel("")
    axRight.autoscale(enable=True, tight=True)  # axis tight
    adjust_spines(axRight, ["bottom"])