Duplicate legend labels with different colors can often result in an error.

Bug report

If one makes a scatter plot (or a line graph) having multiple points with different markers and colors, but the same label, this will result in an error in version 2.1.1 (but not 2.1.0). A small example is provided below:

Code for reproduction

#!/usr/bin/env python
import matplotlib as mpl
mpl.use('Agg')
import matplotlib.pyplot as plt

color='#00ff00ff'
plt.scatter(0.0, 1.0, color=color, marker='o', label='test')
color='#00ffaaff'
plt.scatter(0.5, 0.0, color=color, marker='v', label='test')
plt.legend()
plt.savefig('foo.png')

Actual outcome

This example uses a scatter plot, but the same issue can arise with line plots if one uses a numpy vector as a color (e.g., [0.0, 1.0, 0.0, 1.0]). The error produced for the above example is as follows:

Traceback (most recent call last):
  File "foo.py", line 11, in <module>
    plt.legend()
  File "/home/ryan/.local/lib/python2.7/site-packages/matplotlib/pyplot.py", line 3748, in legend
    ret = gca().legend(*args, **kwargs)
  File "/home/ryan/.local/lib/python2.7/site-packages/matplotlib/axes/_axes.py", line 498, in legend
    **kwargs)
  File "/home/ryan/.local/lib/python2.7/site-packages/matplotlib/legend.py", line 1402, in _parse_legend_args
    handles, labels = _get_legend_handles_labels(axs, handlers)
  File "/home/ryan/.local/lib/python2.7/site-packages/matplotlib/legend.py", line 1360, in _get_legend_handles_labels
    and not _in_handles(handle, label)):
  File "/home/ryan/.local/lib/python2.7/site-packages/matplotlib/legend.py", line 1344, in _in_handles
    if f_h.get_facecolor() != h.get_facecolor():
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

At the end of the day, this ends up being because the face (and edge) colors are converted to a numpy vector, which can't be compared with != since they have multiple elements. With the help of git bisect, I've tracked this down to this commit at the end of October.

Now I'll grant that this only happens if people use the same label for multiple symbols/lines, which is a bad idea...but one can't always prevent end users from doing that.

I suspect that #10031 is related.

Definitely #10031 Thanks a lot for a reproducible example, however, which has eluded us.

This is kind of strange because the same code works fine for ax.plot.

If it's helpful, here's another example (this time using plot()):

#!/usr/bin/env python
import matplotlib as mpl
mpl.use('Agg')
import matplotlib.pyplot as plt
import numpy as np

x = np.arange(0, 10, 0.01)
y0 = np.sin(x)
y1 = np.cos(x)
color = np.array([0.0, 1.0, 0.0, 1.0])
plt.plot(x, y0, color=color, label="test")
color = np.array([0.0, 1.0, 0.6, 1.0])
plt.plot(x, y1, color=color, label="test")
plt.legend()
plt.savefig('foo.png')

Fixed by #10064. Thanks a lot for the bug report!

Navigation Menu

Repository navigation

Duplicate legend labels with different colors can often result in an error. #10053

Bug report

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Duplicate legend labels with different colors can often result in an error. #10053

Description

Bug report

Activity

dpryan79 commented on Dec 20, 2017

jklymak commented on Dec 20, 2017

dpryan79 commented on Dec 21, 2017

jklymak commented on Dec 21, 2017

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions