Empty square for legend for stackplot - python-3.x

Question

I'm trying to generate a stack plot of version data using matplotlib. I have that portion working and displaying properly, but I'm unable to get the legend to display anything other than an empty square in the corner.
ra_ys = np.asarray(ra_ys)
# Going to generate a stack plot of the version stats
fig = plt.figure()
ra_plot = fig.add_subplot(111)
# Our x axis is going to be the dates, but we need them as numbers
x = [date2num(date) for date in dates]
# Plot the data
ra_plot.stackplot(x, ra_ys)
# Setup our legends
ra_plot.legend(ra_versions) #Also tried converting to a tuple
ra_plot.set_title("blah blah words")
print(ra_versions)
# Only want x ticks on the dates we supplied, and want them to display AS dates
ra_plot.set_xticks(x)
ra_plot.set_xticklabels([date.strftime("%m-%d") for date in dates])
plt.show()
ra_ys is a multidimensional array:
[[ 2 2 2 2 2 2 2 2 2 2 1]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 1 1 1 1 1 1 1 1 1 1 1]
[53 52 51 50 50 49 48 48 48 48 47]
[18 19 20 20 20 20 21 21 21 21 21]
[ 0 0 12 15 17 18 19 19 19 19 22]
[ 5 5 3 3 3 3 3 3 3 3 3]
[ 4 4 3 3 2 2 2 2 2 2 2]
[14 14 6 4 3 3 2 2 2 2 2]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 2 2 2 2 2 2 2 2 2 2 2]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 1 1 1 1 1 1 1 1 1 1 1]
[ 3 3 2 2 2 2 2 2 2 2 2]]
x is some dates: [734969.0, 734970.0, 734973.0, 734974.0, 734975.0, 734976.0, 734977.0, 734978.0, 734979.0, 734980.0, 734981.0]
ra_versions is a list: ['4.5.2', '4.5.7', '4.5.8', '5.0.0', '5.0.1', '5.0.10', '5.0.7', '5.0.8', '5.0.9', '5.9.105', '5.9.26', '5.9.27', '5.9.29', '5.9.31', '5.9.32', '5.9.34']
Am I doing something wrong? Can stack plots not have legends?
EDIT: I tried to print the handles and labels for the plot and got two empty lists ([] []):
handles, labels = theplot.get_legend_handles_labels()
print(handles,labels)
I then tested the same figure using the follow code for a proxy handle and it worked. So it looks like the lack of handles is the problem.
p = plt.Rectangle((0, 0), 1, 1, fc="r")
theplot.legend([p], ['test'])
So now the question is, how can I generate a variable number of proxy handles that match the colors of my stack plot?

Answer 1

This is the final (cleaner) approach to getting the legend. Since there are no handles, I generate proxy artists for each line. It's theoretically capable of handling cases where colors are reused, but it'll be confusing.
def plot_version_data(title, dates, versions, version_ys, savename=None):
print("Prepping plot for \"{0}\"".format(title))
fig = plt.figure()
theplot = fig.add_subplot(111)
# Our x axis is going to be the dates, but we need them as numbers
x = [date2num(date) for date in dates]
# Use these colors
colormap = "bgrcmy"
theplot.stackplot(x, version_ys, colors=colormap)
# Make some proxy artists for the legend
p = []
i = 0
for _ in versions:
p.append(plt.Rectangle((0, 0), 1, 1, fc=colormap[i]))
i = (i + 1) % len(colormap)
theplot.legend(p, versions)
theplot.set_ylabel(versions) # Cheating way to handle the legend
theplot.set_title(title)
# Setup the X axis - rotate to keep from overlapping, display like Oct-16,
# make sure there's no random whitespace on either end
plt.xticks(rotation=315)
theplot.set_xticks(x)
theplot.set_xticklabels([date.strftime("%b-%d") for date in dates])
plt.xlim(x[0],x[-1])
if savename:
print("Saving output as \"{0}\"".format(savename))
fig.savefig(os.path.join(sys.path[0], savename))
else:
plt.show()