Skip to content

Commit 1d02261

Browse files
anntzeranntzer
committed
Consolidate agg-buffer examples.
agg_oo_sgskip, webapp_demo_sgskip, and histogram_demo_canvasagg_sgskip were all demonstrating the same thing (how to create a Figure outside of pyplot, attach a canvas, and get the data out either saved to a file or as a string. Consolidate them into a single example, with the minimal amount of plotting. (PIL.Image.frombytes exists since Pillow 2.0 which is the first version to support Py3.) agg_buffer is very similar but uses another (worse?) technique to attach an agg buffer; make data extraction simpler but otherwise keep it separate for now. Other related examples include: - agg_buffer_to_array which directly does `np.asarray(canvas.renderer._renderer)`, which is on the one side bad as it uses a private API, but on the other hand the only way (AFAIK) to access the underlying buffer without copy; - print_stdout_sgskip, which does use pyplot to set up the agg buffer. Let's defer the discussion for these two examples to later.
1 parent 66a6d70 commit 1d02261

File tree

5 files changed

+77
-188
lines changed

5 files changed

+77
-188
lines changed

‎examples/misc/agg_buffer.py

Lines changed: 8 additions & 19 deletions
Original file line numberDiff line numberDiff line change
@@ -9,34 +9,23 @@
99

1010
import numpy as np
1111

12-
import matplotlib.pyplot as plt
1312
from matplotlib.backends.backend_agg import FigureCanvasAgg
14-
15-
16-
try:
17-
from PIL import Image
18-
except ImportError:
19-
raise SystemExit("Pillow must be installed to run this example")
13+
import matplotlib.pyplot as plt
2014

2115
plt.plot([1, 2, 3])
2216

2317
canvas = plt.get_current_fig_manager().canvas
2418

2519
agg = canvas.switch_backends(FigureCanvasAgg)
2620
agg.draw()
27-
s = agg.tostring_rgb()
28-
29-
# get the width and the height to resize the matrix
30-
l, b, w, h = agg.figure.bbox.bounds
31-
w, h = int(w), int(h)
21+
s, (width, height) = agg.print_to_buffer()
3222

33-
X = np.fromstring(s, np.uint8).reshape((h, w, 3))
23+
# Convert to a NumPy array.
24+
X = np.fromstring(s, np.uint8).reshape((height, width, 4))
3425

35-
try:
36-
im = Image.fromstring("RGB", (w, h), s)
37-
except Exception:
38-
im = Image.frombytes("RGB", (w, h), s)
26+
# Pass off to PIL.
27+
from PIL import Image
28+
im = Image.frombytes("RGBA", (width, height), s)
3929

40-
# Uncomment this line to display the image using ImageMagick's
41-
# `display` tool.
30+
# Uncomment this line to display the image using ImageMagick's `display` tool.
4231
# im.show()

‎examples/misc/agg_oo_sgskip.py

Lines changed: 0 additions & 45 deletions
This file was deleted.

‎examples/misc/webapp_demo_sgskip.py

Lines changed: 0 additions & 64 deletions
This file was deleted.

‎examples/user_interfaces/canvasagg_demo.py

Lines changed: 69 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,69 @@
1+
"""
2+
==============
3+
CanvasAgg demo
4+
==============
5+
6+
This example shows how to use the agg backend directly to create images, which
7+
may be of use to web application developers who want full control over their
8+
code without using the pyplot interface to manage figures, figure closing etc.
9+
10+
.. note::
11+
12+
It is not necessary to avoid using the pyplot interface in order to
13+
create figures without a graphical front-end - simply setting
14+
the backend to "Agg" would be sufficient.
15+
16+
In this example, we show how to save the contents of the agg canvas to a file,
17+
and how to extract them to a string, which can in turn be passed off to PIL or
18+
put in a numpy array. The latter functionality allows e.g. to use Matplotlib
19+
inside a cgi-script *without* needing to write a figure to disk.
20+
"""
21+
22+
from matplotlib.backends.backend_agg import FigureCanvasAgg
23+
from matplotlib.figure import Figure
24+
import numpy as np
25+
26+
fig = Figure(figsize=(5, 4), dpi=100)
27+
# A canvas must be manually attached to the figure (pyplot would automatically
28+
# do it). This is done by instantiating the canvas with the figure as
29+
# argument.
30+
canvas = FigureCanvasAgg(fig)
31+
32+
# Do some plotting.
33+
ax = fig.add_subplot(111)
34+
ax.plot([1, 2, 3])
35+
36+
# Option 1: Save the figure to a file; can also be a file-like object (BytesIO,
37+
# etc.).
38+
fig.savefig("test.png")
39+
40+
# Option 2: Save the figure to a string.
41+
canvas.draw()
42+
s, (width, height) = canvas.print_to_buffer()
43+
44+
# Option 2a: Convert to a NumPy array.
45+
X = np.fromstring(s, np.uint8).reshape((height, width, 4))
46+
47+
# Option 2b: Pass off to PIL.
48+
from PIL import Image
49+
im = Image.frombytes("RGBA", (width, height), s)
50+
51+
# Uncomment this line to display the image using ImageMagick's `display` tool.
52+
# im.show()
53+
54+
#############################################################################
55+
#
56+
# ------------
57+
#
58+
# References
59+
# """"""""""
60+
#
61+
# The use of the following functions, methods, classes and modules is shown
62+
# in this example:
63+
64+
import matplotlib
65+
matplotlib.backends.backend_agg.FigureCanvasAgg
66+
matplotlib.figure.Figure
67+
matplotlib.figure.Figure.add_subplot
68+
matplotlib.figure.Figure.savefig
69+
matplotlib.axes.Axes.plot

‎examples/user_interfaces/histogram_demo_canvasagg_sgskip.py

Lines changed: 0 additions & 60 deletions
This file was deleted.

0 commit comments

Comments
 (0)