##// END OF EJS Templates
Reset the interactive namespace __warningregistry__ before executing code...
Reset the interactive namespace __warningregistry__ before executing code Fixes #6611. Idea: Right now, people often don't see important warnings when running code in IPython, because (to a first approximation) any given warning will only issue once per session. Blink and you'll miss it! This is a very common contributor to confused emails to numpy-discussion. E.g.: In [5]: 1 / my_array_with_random_contents /home/njs/.user-python2.7-64bit-3/bin/ipython:1: RuntimeWarning: divide by zero encountered in divide #!/home/njs/.user-python2.7-64bit-3/bin/python Out[5]: array([ 1.77073316, -2.29765021, -2.01800811, ..., 1.13871243, -1.08302964, -8.6185091 ]) Oo, right, guess I gotta be careful of those zeros -- thanks, numpy, for giving me that warning! A few days later: In [592]: 1 / some_other_array Out[592]: array([ 3.07735763, 0.50769289, 0.83984078, ..., -0.67563917, -0.85736257, -1.36511271]) Oops, it turns out that this array had a zero in it too, and that's going to bite me later. But no warning this time! The effect of this commit is to make it so that warnings triggered by the code in cell 5 do *not* suppress warnings triggered by the code in cell 592. Note that this only applies to warnings triggered *directly* by code entered interactively -- if somepkg.foo() calls anotherpkg.bad_func() which issues a warning, then this warning will still only be displayed once, even if multiple cells call somepkg.foo(). But if cell 5 and cell 592 both call anotherpkg.bad_func() directly, then both will get warnings. (Important exception: if foo() is defined *interactively*, and calls anotherpkg.bad_func(), then every cell that calls foo() will display the warning again. This is unavoidable without fixes to CPython upstream.) Explanation: Python's warning system has some weird quirks. By default, it tries to suppress duplicate warnings, where "duplicate" means the same warning message triggered twice by the same line of code. This requires determining which line of code is responsible for triggering a warning, and this is controlled by the stacklevel= argument to warnings.warn. Basically, though, the idea is that if foo() calls bar() which calls baz() which calls some_deprecated_api(), then baz() will get counted as being "responsible", and the warning system will make a note that the usage of some_deprecated_api() inside baz() has already been warned about and doesn't need to be warned about again. So far so good. To accomplish this, obviously, there has to be a record of somewhere which line this was. You might think that this would be done by recording the filename:linenumber pair in a dict inside the warnings module, or something like that. You would be wrong. What actually happens is that the warnings module will use stack introspection to reach into baz()'s execution environment, create a global (module-level) variable there named __warningregistry__, and then, inside this dictionary, record just the line number. Basically, it assumes that any given module contains only one line 1, only one line 2, etc., so storing the filename is irrelevant. Obviously for interactive code this is totally wrong -- all cells share the same execution environment and global namespace, and they all contain a new line 1. Currently the warnings module treats these as if they were all the same line. In fact they are not the same line; once we have executed a given chunk of code, we will never see those particular lines again. As soon as a given chunk of code finishes executing, its line number labels become meaningless, and the corresponding warning registry entries become meaningless as well. Therefore, with this patch we delete the __warningregistry__ each time we execute a new block of code.

File last commit:

r18091:db4cc974
r18548:61431d7d
Show More
pylabtools.py
368 lines | 12.5 KiB | text/x-python | PythonLexer
# -*- coding: utf-8 -*-
"""Pylab (matplotlib) support utilities."""
from __future__ import print_function
# Copyright (c) IPython Development Team.
# Distributed under the terms of the Modified BSD License.
from io import BytesIO
from IPython.core.display import _pngxy
from IPython.utils.decorators import flag_calls
from IPython.utils import py3compat
# If user specifies a GUI, that dictates the backend, otherwise we read the
# user's mpl default from the mpl rc structure
backends = {'tk': 'TkAgg',
'gtk': 'GTKAgg',
'gtk3': 'GTK3Agg',
'wx': 'WXAgg',
'qt': 'Qt4Agg', # qt3 not supported
'qt4': 'Qt4Agg',
'qt5': 'Qt5Agg',
'osx': 'MacOSX',
'nbagg': 'nbAgg',
'inline' : 'module://IPython.kernel.zmq.pylab.backend_inline'}
# We also need a reverse backends2guis mapping that will properly choose which
# GUI support to activate based on the desired matplotlib backend. For the
# most part it's just a reverse of the above dict, but we also need to add a
# few others that map to the same GUI manually:
backend2gui = dict(zip(backends.values(), backends.keys()))
# Our tests expect backend2gui to just return 'qt'
backend2gui['Qt4Agg'] = 'qt'
# In the reverse mapping, there are a few extra valid matplotlib backends that
# map to the same GUI support
backend2gui['GTK'] = backend2gui['GTKCairo'] = 'gtk'
backend2gui['GTK3Cairo'] = 'gtk3'
backend2gui['WX'] = 'wx'
backend2gui['CocoaAgg'] = 'osx'
#-----------------------------------------------------------------------------
# Matplotlib utilities
#-----------------------------------------------------------------------------
def getfigs(*fig_nums):
"""Get a list of matplotlib figures by figure numbers.
If no arguments are given, all available figures are returned. If the
argument list contains references to invalid figures, a warning is printed
but the function continues pasting further figures.
Parameters
----------
figs : tuple
A tuple of ints giving the figure numbers of the figures to return.
"""
from matplotlib._pylab_helpers import Gcf
if not fig_nums:
fig_managers = Gcf.get_all_fig_managers()
return [fm.canvas.figure for fm in fig_managers]
else:
figs = []
for num in fig_nums:
f = Gcf.figs.get(num)
if f is None:
print('Warning: figure %s not available.' % num)
else:
figs.append(f.canvas.figure)
return figs
def figsize(sizex, sizey):
"""Set the default figure size to be [sizex, sizey].
This is just an easy to remember, convenience wrapper that sets::
matplotlib.rcParams['figure.figsize'] = [sizex, sizey]
"""
import matplotlib
matplotlib.rcParams['figure.figsize'] = [sizex, sizey]
def print_figure(fig, fmt='png', bbox_inches='tight', **kwargs):
"""Print a figure to an image, and return the resulting file data
Returned data will be bytes unless ``fmt='svg'``,
in which case it will be unicode.
Any keyword args are passed to fig.canvas.print_figure,
such as ``quality`` or ``bbox_inches``.
"""
from matplotlib import rcParams
# When there's an empty figure, we shouldn't return anything, otherwise we
# get big blank areas in the qt console.
if not fig.axes and not fig.lines:
return
dpi = rcParams['savefig.dpi']
if fmt == 'retina':
dpi = dpi * 2
fmt = 'png'
# build keyword args
kw = dict(
format=fmt,
facecolor=fig.get_facecolor(),
edgecolor=fig.get_edgecolor(),
dpi=dpi,
bbox_inches=bbox_inches,
)
# **kwargs get higher priority
kw.update(kwargs)
bytes_io = BytesIO()
fig.canvas.print_figure(bytes_io, **kw)
data = bytes_io.getvalue()
if fmt == 'svg':
data = data.decode('utf-8')
return data
def retina_figure(fig, **kwargs):
"""format a figure as a pixel-doubled (retina) PNG"""
pngdata = print_figure(fig, fmt='retina', **kwargs)
w, h = _pngxy(pngdata)
metadata = dict(width=w//2, height=h//2)
return pngdata, metadata
# We need a little factory function here to create the closure where
# safe_execfile can live.
def mpl_runner(safe_execfile):
"""Factory to return a matplotlib-enabled runner for %run.
Parameters
----------
safe_execfile : function
This must be a function with the same interface as the
:meth:`safe_execfile` method of IPython.
Returns
-------
A function suitable for use as the ``runner`` argument of the %run magic
function.
"""
def mpl_execfile(fname,*where,**kw):
"""matplotlib-aware wrapper around safe_execfile.
Its interface is identical to that of the :func:`execfile` builtin.
This is ultimately a call to execfile(), but wrapped in safeties to
properly handle interactive rendering."""
import matplotlib
import matplotlib.pylab as pylab
#print '*** Matplotlib runner ***' # dbg
# turn off rendering until end of script
is_interactive = matplotlib.rcParams['interactive']
matplotlib.interactive(False)
safe_execfile(fname,*where,**kw)
matplotlib.interactive(is_interactive)
# make rendering call now, if the user tried to do it
if pylab.draw_if_interactive.called:
pylab.draw()
pylab.draw_if_interactive.called = False
return mpl_execfile
def select_figure_formats(shell, formats, **kwargs):
"""Select figure formats for the inline backend.
Parameters
==========
shell : InteractiveShell
The main IPython instance.
formats : str or set
One or a set of figure formats to enable: 'png', 'retina', 'jpeg', 'svg', 'pdf'.
**kwargs : any
Extra keyword arguments to be passed to fig.canvas.print_figure.
"""
from matplotlib.figure import Figure
from IPython.kernel.zmq.pylab import backend_inline
svg_formatter = shell.display_formatter.formatters['image/svg+xml']
png_formatter = shell.display_formatter.formatters['image/png']
jpg_formatter = shell.display_formatter.formatters['image/jpeg']
pdf_formatter = shell.display_formatter.formatters['application/pdf']
if isinstance(formats, py3compat.string_types):
formats = {formats}
# cast in case of list / tuple
formats = set(formats)
[ f.pop(Figure, None) for f in shell.display_formatter.formatters.values() ]
supported = {'png', 'png2x', 'retina', 'jpg', 'jpeg', 'svg', 'pdf'}
bad = formats.difference(supported)
if bad:
bs = "%s" % ','.join([repr(f) for f in bad])
gs = "%s" % ','.join([repr(f) for f in supported])
raise ValueError("supported formats are: %s not %s" % (gs, bs))
if 'png' in formats:
png_formatter.for_type(Figure, lambda fig: print_figure(fig, 'png', **kwargs))
if 'retina' in formats or 'png2x' in formats:
png_formatter.for_type(Figure, lambda fig: retina_figure(fig, **kwargs))
if 'jpg' in formats or 'jpeg' in formats:
jpg_formatter.for_type(Figure, lambda fig: print_figure(fig, 'jpg', **kwargs))
if 'svg' in formats:
svg_formatter.for_type(Figure, lambda fig: print_figure(fig, 'svg', **kwargs))
if 'pdf' in formats:
pdf_formatter.for_type(Figure, lambda fig: print_figure(fig, 'pdf', **kwargs))
#-----------------------------------------------------------------------------
# Code for initializing matplotlib and importing pylab
#-----------------------------------------------------------------------------
def find_gui_and_backend(gui=None, gui_select=None):
"""Given a gui string return the gui and mpl backend.
Parameters
----------
gui : str
Can be one of ('tk','gtk','wx','qt','qt4','inline').
gui_select : str
Can be one of ('tk','gtk','wx','qt','qt4','inline').
This is any gui already selected by the shell.
Returns
-------
A tuple of (gui, backend) where backend is one of ('TkAgg','GTKAgg',
'WXAgg','Qt4Agg','module://IPython.kernel.zmq.pylab.backend_inline').
"""
import matplotlib
if gui and gui != 'auto':
# select backend based on requested gui
backend = backends[gui]
else:
# We need to read the backend from the original data structure, *not*
# from mpl.rcParams, since a prior invocation of %matplotlib may have
# overwritten that.
# WARNING: this assumes matplotlib 1.1 or newer!!
backend = matplotlib.rcParamsOrig['backend']
# In this case, we need to find what the appropriate gui selection call
# should be for IPython, so we can activate inputhook accordingly
gui = backend2gui.get(backend, None)
# If we have already had a gui active, we need it and inline are the
# ones allowed.
if gui_select and gui != gui_select:
gui = gui_select
backend = backends[gui]
return gui, backend
def activate_matplotlib(backend):
"""Activate the given backend and set interactive to True."""
import matplotlib
matplotlib.interactive(True)
# Matplotlib had a bug where even switch_backend could not force
# the rcParam to update. This needs to be set *before* the module
# magic of switch_backend().
matplotlib.rcParams['backend'] = backend
import matplotlib.pyplot
matplotlib.pyplot.switch_backend(backend)
# This must be imported last in the matplotlib series, after
# backend/interactivity choices have been made
import matplotlib.pylab as pylab
pylab.show._needmain = False
# We need to detect at runtime whether show() is called by the user.
# For this, we wrap it into a decorator which adds a 'called' flag.
pylab.draw_if_interactive = flag_calls(pylab.draw_if_interactive)
def import_pylab(user_ns, import_all=True):
"""Populate the namespace with pylab-related values.
Imports matplotlib, pylab, numpy, and everything from pylab and numpy.
Also imports a few names from IPython (figsize, display, getfigs)
"""
# Import numpy as np/pyplot as plt are conventions we're trying to
# somewhat standardize on. Making them available to users by default
# will greatly help this.
s = ("import numpy\n"
"import matplotlib\n"
"from matplotlib import pylab, mlab, pyplot\n"
"np = numpy\n"
"plt = pyplot\n"
)
exec(s, user_ns)
if import_all:
s = ("from matplotlib.pylab import *\n"
"from numpy import *\n")
exec(s, user_ns)
# IPython symbols to add
user_ns['figsize'] = figsize
from IPython.core.display import display
# Add display and getfigs to the user's namespace
user_ns['display'] = display
user_ns['getfigs'] = getfigs
def configure_inline_support(shell, backend):
"""Configure an IPython shell object for matplotlib use.
Parameters
----------
shell : InteractiveShell instance
backend : matplotlib backend
"""
# If using our svg payload backend, register the post-execution
# function that will pick up the results for display. This can only be
# done with access to the real shell object.
# Note: if we can't load the inline backend, then there's no point
# continuing (such as in terminal-only shells in environments without
# zeromq available).
try:
from IPython.kernel.zmq.pylab.backend_inline import InlineBackend
except ImportError:
return
from matplotlib import pyplot
cfg = InlineBackend.instance(parent=shell)
cfg.shell = shell
if cfg not in shell.configurables:
shell.configurables.append(cfg)
if backend == backends['inline']:
from IPython.kernel.zmq.pylab.backend_inline import flush_figures
shell.events.register('post_execute', flush_figures)
# Save rcParams that will be overwrittern
shell._saved_rcParams = dict()
for k in cfg.rc:
shell._saved_rcParams[k] = pyplot.rcParams[k]
# load inline_rc
pyplot.rcParams.update(cfg.rc)
else:
from IPython.kernel.zmq.pylab.backend_inline import flush_figures
try:
shell.events.unregister('post_execute', flush_figures)
except ValueError:
pass
if hasattr(shell, '_saved_rcParams'):
pyplot.rcParams.update(shell._saved_rcParams)
del shell._saved_rcParams
# Setup the default figure format
select_figure_formats(shell, cfg.figure_formats, **cfg.print_figure_kwargs)