Jun 10, 2024 · Jun 11, 2024
diff --git a/galleries/examples/misc/combining_transforms.py b/galleries/examples/misc/combining_transforms.py
 """
 =========================
 Combining Transformations
 =========================

 This example showcases how to combine transformations, including Affine
 and blended transformations.

 The goal of this exercise is to plot some circles with correct aspect
 ratios on axes that are unequal. The interesting thing is that the
 circles have a fixed size along the y-axis in data space rather than
 axes, figure or display space.

 To ensure a large disparity in the data axis scales, we plot
 categorical vs time data. The x-axis is set to a short time interval of
 a few minutes. The y-axis is integers corresponding to the category
 index.

 The transform graph is constructed as follows:

 - A blended transform is made to scale the data:

  - The y direction is the scale-only portion of ``ax.transData``
    obtained with ``AffineDeltaTransform``
  - The x direction is a reflection of the y, made using an ``Affine2D``
    reflection matrix

 - The blended transformation is added to a ``ScaledTranslation`` to place
  the circles correctly

 As a secondary showcase, this example shows how to work with datetimes
 along one axis when constructing the rectangular patches behind the
 circles. It also demonstrates a way to increment the line color cycler.
 """

 from datetime import datetime

 import numpy as np

 from matplotlib import dates as mdate
 from matplotlib import patches as mpatch
 from matplotlib import pyplot as plt
 from matplotlib import transforms as mtrans

 data = {
    "A": [datetime(2024, 4, 10, 3, 10, 22),
          datetime(2024, 4, 10, 3, 21, 13),
          datetime(2024, 4, 10, 3, 25, 41)],
    "B": [datetime(2024, 4, 10, 3, 15, 55),
          datetime(2024, 4, 10, 3, 40, 8)],
    "C": [datetime(2024, 4, 10, 3, 12, 18),
          datetime(2024, 4, 10, 3, 23, 32),
          datetime(2024, 4, 10, 3, 32, 12)],
 }

 fig, ax = plt.subplots(constrained_layout=True)
 ax.invert_yaxis()

 # The scaling transforms only need to be set up once
 vertical_scale_transform = mtrans.AffineDeltaTransform(ax.transData)
 reflection = mtrans.Affine2D.from_values(0, 1, 1, 0, 0, 0)
 uniform_scale_transform = mtrans.blended_transform_factory(
    reflection + vertical_scale_transform + reflection, vertical_scale_transform)

 # Draw some rectangle spanning each dataset
 for i, dates in enumerate(data.values()):
    start = mdate.date2num(dates[0])
    end = mdate.date2num(dates[-1])
    width = end - start
    color = ax._get_lines.get_next_color()
    ax.add_patch(mpatch.Rectangle((start, i - 0.4), width, 0.8, color=color))

    # Draw a circle at each event
    for event in dates:
        x = mdate.date2num(event)
        t = uniform_scale_transform + mtrans.ScaledTranslation(x, i, ax.transData)
        ax.add_patch(mpatch.Circle((0, 0), 0.2, facecolor="w", edgecolor="k",
                                   linewidth=2, transform=t))

 # Set the y-axis to show the data labels
 ax.set_yticks(np.arange(len(data)))
 ax.set_yticklabels(list(data))

 # Set the x-axis to display datetimes
 ax.xaxis.set_major_locator(locator := mdate.AutoDateLocator())
 ax.xaxis.set_major_formatter(mdate.AutoDateFormatter(locator))

 ax.autoscale_view()

 plt.show()
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,90 @@
		"""
		=========================
		Combining Transformations
		=========================

		This example showcases how to combine transformations, including Affine
		and blended transformations.

		The goal of this exercise is to plot some circles with correct aspect
		ratios on axes that are unequal. The interesting thing is that the
		circles have a fixed size along the y-axis in data space rather than
		axes, figure or display space.

		To ensure a large disparity in the data axis scales, we plot
		categorical vs time data. The x-axis is set to a short time interval of
		a few minutes. The y-axis is integers corresponding to the category
		index.

		The transform graph is constructed as follows:

		- A blended transform is made to scale the data:

		- The y direction is the scale-only portion of ``ax.transData``
		obtained with ``AffineDeltaTransform``
		- The x direction is a reflection of the y, made using an ``Affine2D``
		reflection matrix

		- The blended transformation is added to a ``ScaledTranslation`` to place
		the circles correctly

		As a secondary showcase, this example shows how to work with datetimes
		along one axis when constructing the rectangular patches behind the
		circles. It also demonstrates a way to increment the line color cycler.
		"""

		from datetime import datetime

		import numpy as np

		from matplotlib import dates as mdate
		from matplotlib import patches as mpatch
		from matplotlib import pyplot as plt
		from matplotlib import transforms as mtrans

		data = {
		"A": [datetime(2024, 4, 10, 3, 10, 22),
		datetime(2024, 4, 10, 3, 21, 13),
		datetime(2024, 4, 10, 3, 25, 41)],
		"B": [datetime(2024, 4, 10, 3, 15, 55),
		datetime(2024, 4, 10, 3, 40, 8)],
		"C": [datetime(2024, 4, 10, 3, 12, 18),
		datetime(2024, 4, 10, 3, 23, 32),
		datetime(2024, 4, 10, 3, 32, 12)],
		}

		fig, ax = plt.subplots(constrained_layout=True)
		ax.invert_yaxis()

		# The scaling transforms only need to be set up once
		vertical_scale_transform = mtrans.AffineDeltaTransform(ax.transData)
		reflection = mtrans.Affine2D.from_values(0, 1, 1, 0, 0, 0)
		uniform_scale_transform = mtrans.blended_transform_factory(
		reflection + vertical_scale_transform + reflection, vertical_scale_transform)

		# Draw some rectangle spanning each dataset
		for i, dates in enumerate(data.values()):
		start = mdate.date2num(dates[0])
		end = mdate.date2num(dates[-1])
		width = end - start
		color = ax._get_lines.get_next_color()
		ax.add_patch(mpatch.Rectangle((start, i - 0.4), width, 0.8, color=color))

		# Draw a circle at each event
		for event in dates:
		x = mdate.date2num(event)
		t = uniform_scale_transform + mtrans.ScaledTranslation(x, i, ax.transData)
		ax.add_patch(mpatch.Circle((0, 0), 0.2, facecolor="w", edgecolor="k",
		linewidth=2, transform=t))

		# Set the y-axis to show the data labels
		ax.set_yticks(np.arange(len(data)))
		ax.set_yticklabels(list(data))

		# Set the x-axis to display datetimes
		ax.xaxis.set_major_locator(locator := mdate.AutoDateLocator())
		ax.xaxis.set_major_formatter(mdate.AutoDateFormatter(locator))

		ax.autoscale_view()

		plt.show()