12

I have a follow up question to this question.

Is it possible to streamline the figure generation by having multiple python scripts that work on different parts of the figure?

For example, if I have the following functions:

FunctionA: Draw a histogram of something
FunctionB: Draw a box with a text in it
FunctionC: Draw a plot of something C
FunctionD: Draw a plot of something D

How do I go about reusing the above functions in different scripts? If I wanted, for instance, to create a figure with a histogram with a plot of something C, I would somehow call FunctionA and FunctionC from my script. Or, if I wanted a figure with the two plots, I'd call FunctionC and FunctionD.

I'm not sure if I'm explaining myself clearly, but another way of asking this question is this: how do I pass a figure object to a function and then have the function draw something to the passed figure object and then return it back to the main script to add other things like the title or something?

Community
  • 1
  • 1
aspade
  • 575
  • 1
  • 6
  • 10

2 Answers2

8

Here you want to use the Artist objects, and pass them as needed to the functions:

import numpy as np
import matplotlib.pyplot as plt

def myhist(ax, color):
    ax.hist(np.log(np.arange(1, 10, .1)), facecolor=color)

def say_something(ax, words):
    t = ax.text(.2, 20., words)
    make_a_dim_yellow_bbox(t)

def make_a_dim_yellow_bbox(txt):
    txt.set_bbox(dict(facecolor='yellow', alpha=.2))

fig = plt.figure()
ax0 = fig.add_subplot(1,2,1)
ax1 = fig.add_subplot(1,2,2)

myhist(ax0, 'blue')
myhist(ax1, 'green')

say_something(ax0, 'this is the blue plot')
say_something(ax1, 'this is the green plot')

plt.show()

alt text

Glorfindel
  • 21,988
  • 13
  • 81
  • 109
tom10
  • 67,082
  • 10
  • 127
  • 137
  • You're welcome! To format the code, indent the whole code block by four extra spaces, and use spaces instead of tabs within your code. There's more on formatting here: http://stackoverflow.com/editing-help – tom10 Sep 12 '09 at 03:07
  • Why don't `myhist()` and `say_something()` have to return `ax` back into the outer namespace? – crypdick Nov 11 '15 at 05:03
  • 1
    @RovingRichard: Python passes by reference, so the two fcns act on the objects that are passed into them, which are the same objects (eg, have same `id`) as in the global namespace. (I hope this answers your question, but if it doesn't, you'd need to say why you think the fcns would have to return `ax`.) – tom10 Nov 12 '15 at 14:10
  • @tom10 D'oh! This whole time I thought when I pass `object1` to a function, the function created it's own namespace with a copy `object2`, and that if I wanted to update the `object1` I had to `return object2`. – crypdick Nov 12 '15 at 21:22
0

Okey, I've figured out how to do this. It was a lot simpler than what I had imagined. It just required a bit of reading here with the figure and axes classes.

In your main script:

import pylab as plt  
import DrawFns  
fig = plt.figure()  
(do something with fig)  
DrawFns.WriteText(fig, 'Testing')  
plt.show()

In your DrawFns.py: 

def WriteText(_fig, _text):  
[indent]_fig.text(0, 0, _text)

And that's it! And I can add more functions in DrawFns.py and call them from any script as long as they are included with import call. :D

Nathan Fellman
  • 122,701
  • 101
  • 260
  • 319
aspade
  • 575
  • 1
  • 6
  • 10
  • This is the right idea, except generally the figure contains one or more axes and most of the action is in calling the axes methods to make your plots, so it's easier to make the figure, then the axes, and pass the axes (or some object within them) to the functions. See my answer for an example. – tom10 Sep 12 '09 at 01:02
  • Sorry, I just saw this. Thanks for the tip. I'll change the code accordingly (offline). Thank you! – aspade Sep 12 '09 at 02:45