At some point in the last few days, Matplotlib stopped working for me on OS X. Here's the error I get when trying to import matplotlib:
Traceback (most recent call last):
File "/my/path/to/script/my_script.py", line 15, in <module>
import matplotlib.pyplot as plt
File "/Library/Python/2.7/site-packages/matplotlib/pyplot.py", line 34, in <module>
from matplotlib.figure import Figure, figaspect
File "/Library/Python/2.7/site-packages/matplotlib/figure.py", line 40, in <module>
from matplotlib.axes import Axes, SubplotBase, subplot_class_factory
File "/Library/Python/2.7/site-packages/matplotlib/axes/__init__.py", line 4, in <module>
from ._subplots import *
File "/Library/Python/2.7/site-packages/matplotlib/axes/_subplots.py", line 10, in <module>
from matplotlib.axes._axes import Axes
File "/Library/Python/2.7/site-packages/matplotlib/axes/_axes.py", line 22, in <module>
import matplotlib.dates as _ # <-registers a date unit converter
File "/Library/Python/2.7/site-packages/matplotlib/dates.py", line 126, in <module>
from dateutil.rrule import (rrule, MO, TU, WE, TH, FR, SA, SU, YEARLY,
File "/Library/Python/2.7/site-packages/dateutil/rrule.py", line 14, in <module>
from six.moves import _thread
ImportError: cannot import name _thread
The only system change I can think of was the Apple-forced NTP update and maybe some permission changes I did in /usr/local to get Brew working again.
I tried reinstalling both Matplotlib and Python-dateutil via Pip, but this did not help. Also tried a reboot. I'm running Python 2.7.6, which is located in /usr/bin/python. I'm running Yosemite (OS X 10.10.1).
sudo pip uninstall python-dateutil
sudo pip install python-dateutil==2.2
I had the same error message this afternoon as well, although I did recently upgrade to Yosemite. I'm not totally sure I understand why reverting dateutil to a previous version works for me, but since running the above I'm having no trouble (I generally use pyplot inline in an ipython notebook).
This problem is fixed in the latest six and dateutil versions. However, in OS X, even if you update your six to the latest version, you might not actually update it correctly. This is what happened to me:
After doing a pip2 install six -U, the new six module was installed in /Library/Python/2.7/site-packages/. However, when I loaded six in a python 2.7 terminal, and checked its path, this is what I got:
import six
print six.__file__
/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/six.pyc
So, python was using an old version of six, which I removed by typing:
rm -rf /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/six.*
This fixed this issue for me.
Installing the python-dateutil==2.2 did not work for me.
But a quick-and-dirty workaround did work! I replace six.py in python 2.7 with the six.py from python 3.4 (virtualenv). Since, I have the problem in 2.7 but not 3.4.
UPDATE
I had the same problem again after reinstalling python (and after upgrading to El Capitan). Un-obvious thing is that this error occurs only in the IPython shell and notebook (when I do import matplotlib.pyplot as plt) but works fine from a Python shell.
So a better solution (that did work in my case) without a dirty work-around is to force install both six and ipython. Here is what I did to have this fixed :
$ pip install --ignore-installed six
$ pip install --ignore-installed ipython
It is possible that you have a perfectly installed version of any packages you have installed, but the version used by default is not the one you want. You can see the list of paths that python search from in order to find its packages as follows:
>>> import sys
>>> sys.path
In order to let python search first the most updated version of certain package, instead of removing the system version, what can be done is to set the system variable PYTHONPATH in the ~/.bash_profile (or ~/.bashrc if linux) config file to the path where the new packages are installed:
export PYTHONPATH=/Library/Python/2.7/site-packages
An alternative is to modify the python path inside your python script by adding the path at the beginning of the path list:
import sys
sys.path.insert(1,'/Library/Python/2.7/site-packages')
This needs to be done for every script you need a certain package version. You might want for some reason use an older version that you have installed.
BTW all my installations with easy_install, or pip, or from sources go to /Library/Python/2.7/site-packages
This worked en EL Capitan, and now also in macOS Sierra (10.12.2)
Related
I've used the following tutorial to install Python, numpy, scipy and matplotlib:
https://penandpants.com/2012/03/01/install-python-2/
I downloaded each and installed by double-clicking. I'm using a Mac with OS X Yosemite Version 10.10.5
When I try to import numpy, scipy and matplotlib in IDLE I get the follow error message:
`>>> import numpy
Traceback (most recent call last):
File "<pyshell#2>", line 1, in <module>
import numpy
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/__init__.py", line 137, in <module>
import add_newdocs
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/add_newdocs.py", line 9, in <module>
from numpy.lib import add_newdoc
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/lib/__init__.py", line 4, in <module>
from type_check import *
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/lib/type_check.py", line 8, in <module>
import numpy.core.numeric as _nx
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/core/__init__.py", line 5, in <module>
import multiarray
ImportError: dlopen(/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/core/multiarray.so, 2): no suitable image found. Did find:
/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/core/multiarray.so: no matching architecture in universal wrapper`
Can anyone please explain to me how to resolve this?
Thanks
use terminal and type
sudo pip install numpy
sudo pip install matplotlib
sudo pip install scipy
for python3+ use pip3
I can't say for sure what's wrong with your installation, but I suspect it's due to a some PATH-conusion... Python 2.7 and Numpy should be installed on your Mac by default.
If you run the command which python, you'll see which executable the python-commands invokes... is it the same as the one you installed?
If you write python in the terminal, and press tab twice, do you see multiple options of installed python versions?
Alternative method of installation
I prefer the Anaconda Python Distribution, which you can download like the full version or minimal version, which is the one I prefer.
If you install this, you should consider uninstalling the one you just installed.
Here's how to install the mini version:
Download the installer, and run it by opening a terminal in the folder with the installer, and running the command bash Miniconda2-latest-MacOSX-x86_64.sh. This will start the text-based installer, and if you read the instructions on screen, you'll be fine.
Notice that you have to scroll to the end of the licence agreement, before you can write yes when asked to accept the terms stated.
Say yes when it asked you if you'd like it to add Anaconda Python to your path.
Anaconda Python is different from other python distributions in the regard that it comes with it's own package and environment manager conda. With conda you can have multiple versions of Python installed at the same time (or several environments of the same Python version, but with different packages or versions of packages).
To create a new environment with the name py27, which is using Python 2.7, write conda create -n py27 python=2.7 --yes. Once created, activate the environment by source activate py27.
With the new environment now active, install the packages you need by using conda: conda install scipy matplotlib --yes.
I'd also recommend installing Ipython, which is an improved shell, and possibly the Jupyter notebook: conda install ipython --yes and conda install notebook --yes.
Hope it helps.
I first came across the answer to this question where I found out I needed to install my own backend framework. Since the answer mentioned PyQt4, I chose to go with that. Following links in the doc, I eventually ended up downloaded SIP (pre-req for PyQt4) and then PyQt4 from here. Finally, in my code, I have:
import matplotlib
matplotlib.use('Qt4agg') # need to call use() before importing plt
import matplotlib.pyplot as plt
However I'm still getting this error:
Traceback (most recent call last):
File ".../venv/lib/python3.5/site-packages/matplotlib/backends/qt_compat.py", line 159, in
from PySide import QtCore, QtGui, version, version_info
ImportError: No module named 'PySide'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File ".../program.py", line 7, in
import matplotlib.pyplot as plt File ".../venv/lib/python3.5/site-packages/matplotlib/pyplot.py", line 114,
in
_backend_mod, new_figure_manager, draw_if_interactive, _show = pylab_setup()
File
".../venv/lib/python3.5/site-packages/matplotlib/backends/init.py",
line 32, in pylab_setup
globals(),locals(),[backend_name],0)
File
".../venv/lib/python3.5/site-packages/matplotlib/backends/backend_qt4agg.py",
line 18, in
from .backend_qt5agg import FigureCanvasQTAggBase as _FigureCanvasQTAggBase
File
".../venv/lib/python3.5/site-packages/matplotlib/backends/backend_qt5agg.py",
line 15, in
from .backend_qt5 import QtCore
File
".../venv/lib/python3.5/site-packages/matplotlib/backends/backend_qt5.py",
line 31, in
from .qt_compat import QtCore, QtGui, QtWidgets, _getSaveFileName, version
File
".../venv/lib/python3.5/site-packages/matplotlib/backends/qt_compat.py",
line 162, in
"Matplotlib qt-based backends require an external PyQt4, PyQt5,\n" ImportError: Matplotlib qt-based backends require an external PyQt4,
PyQt5, or PySide package to be installed, but it was not found.
Has anyone experienced this before? Any debugging advice or help on where to go from here?
EDIT: I'll add that trying to import PyQt4 from my virtual environment works, so I'm not sure why matplotlib isn't finding it...
EDIT2: Not sure if it matters but I'm using PyCharm
I struggled with this for days and have finally arrived at a pretty simple solution after looking through loads of different stack overflow posts.
This is my solution for MacOSX and Python3.X, for someone who has already installed python3 via homebrew (i.e. brew install python3) and has virtualenv installed and a virtual environment already created with python3 (e.g. via virtualenvwrapper, mkvirtualenv myvenv -p python3):
(1) Install pyqt and sip using brew:
$ brew install sip --with-python3
$ brew install pyqt --with-python3
(2) Link the pyqt and sip files installed in /usr/local/Cellar/ (the default location for homebrew installations) to the site-packages directory in your virtualenv:
$ ln -s /usr/local/Cellar/sip/{SIPVERSION}/lib/python3.X/site-packages/*.* ~/{VIRTUALENVHOME}/{VIRTUALENVNAME}/lib/python3.X/site-packages/
$ ln -s /usr/local/Cellar/pyqt/{PYQTVERSION}/lib/python3.X/site-packages/PyQt4/*.* ~/{VIRTUALENVHOME}/{VIRTUALENVNAME}/lib/python3.X/site-packages/PyQt4
Make sure to modify the text in curly brackets {} as needed for your system, where VIRTUALENVHOME is the path to your virtual environment, VIRTUALENVNAME is the name of it, and SIPVERSION and PYQTVERSION are the versions of sip and pyqt that you downloaded. These can be found by simply looking in their corresponding directories /usr/local/Cellar/sip and /usr/local/Cellar/pyqt (the contents of these should be a single directory with the version number). Also don't forget to plug in the version of python3 you are using! (into all places where it says python3.X)
(3) pip install matplotlib in your virtualenv (if you haven't already) and modify the matplotlibrc so that matplotlib uses the PyQt4 backend instead of the default macosx backend:
# activate your virtual environment
$ workon VIRTUALENVNAME # if you are using virtualenvwrapper
$ #source ~/{VIRTUALENVHOME}/{VIRTUALENVNAME}/bin/activate # if you are not using virtualenvwrapper
# if you haven't already, install matplotlib
$ pip install matplotlib
# modify the matplotlibrc file to change the backend it is using
$ nano ~/{VIRTUALENVHOME}/{VIRTUALENVNAME}/lib/python3.5/site-packages/matplotlib/mpl-data/matplotlibrc
This will open matplotlibrc in the default terminal text editor (alternatively you can use open -e instead of nano to open it in TextEdit). Go down to the first section after #### CONFIGURATION BEGINS HERE and change the line backend : macosx to backend : Qt4Agg. In the next section, uncomment the line backend.qt4 : PyQt4.
With this done, you should be up and running. You can test whether this worked with the following simple lines of code (assuming you have ipython installed in your virtual environment):
$ ipython -pylab
>>> import numpy as np
>>> plot(np.arange(10))
A plot should appear with a straight line. If this doesn't work, try reinstalling matplotlib in your virtual environment (i.e. pip uninstall matplotlib and pip install matplotlib).
Resolved this issue following this:
pip install matplotlib
There is a directory in you root called ~/.matplotlib.
Create a file ~/.matplotlib/matplotlibrc there and add the following code: backend: TkAgg
Have you though about using conda instead of virtualenv?
conda create -n matplotenv matplotlib
source activate matplotenv
python -c "import matplotlib.pyplot as plt
p = plt.plot([1,2,3])
plt.show()"
source deactivate
I had the same problem, and what I did actually solved it (it might help if you didn't understand everything Jorge said!) :
1. go on anaconda navigator and launch qtconsole
2. type :$ brew install sip --with-python3
$ brew install pyqt --with-python3
3. copy the result of it in python, and execute.
Now try making a graph, it should work!
I am a Macintosh Yosemite user. When I try to import matplotlib I get the following error.
import matplotlib
Traceback (most recent call last):
File "<pyshell#11>", line 1, in <module>
import matplotlib
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/matplotlib/__init__.py", line 180, in <module>
from matplotlib.cbook import is_string_like
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/matplotlib/cbook.py", line 33, in <module>
import numpy as np
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/numpy/__init__.py", line 170, in <module>
from . import add_newdocs
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/numpy/add_newdocs.py", line 13, in <module>
from numpy.lib import add_newdoc
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/numpy/lib/__init__.py", line 8, in <module>
from .type_check import *
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/numpy/lib/type_check.py", line 11, in <module>
import numpy.core.numeric as _nx
File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/numpy/core/__init__.py", line 6, in <module>
from . import multiarray
ImportError: cannot import name 'multiarray'
When I use from pylab import axis that would work just fine. I do not know what is going on and I am just so lost. I have tried using homebrew, macports installs, dmg installs. I also have very little experience installing things through terminal so I just followed what other people said. Though, that is still not working.
Mac OS X ships with python by default. And there is one provided by homebrew. I would recommend using homebrew python over the default python.
Here, I suspect your numpy installation has landed up in the site-packages directory managed by the non-homebrew pip package manager whereas matplotlib package is installed in a different site-packages directory. (But I am not sure). Nevertheless it has something to do with multiple python's / package managers being used. This may not be the best answer, but so far the only solution I can think of to fix your problems is to uninstall pip and also remove everything in any python site-packages directory you can find on your filesystem,. And install python via homebrew again, and then install all packages required using the pip (which gets installed automatically when you install python using homebrew)
Warning: Make sure you list out the package names and store the names somewhere before deleting them because you will have to install them again.
brew uninstall python
#(ATTN) Uninstall macports and don't use it with brew
#(ATTN) Delete the contents of all python site-packages directories
rm /usr/local/lib/python2.7/site-packages/*
brew install python
# Homebrew comes with its own pip installed
pip install <package1>, <package2> ...
Here is a related question from someone who faced a similar problem:
Numpy build fails with cannot import multiarray
My advice: Don't use Homebrew and Macports or any other package manager together. They mess up with each other and I have faced the consequences in the past.I just use homebrew python now. For installing scientific python packages either Anaconda or Canopy (choose one) are really helpful, which can be installed on top of the homebrew python.
After having a lot of problems similar to the ones you describe, using sudo pip install -U matplotlib worked fine for me.
I've installed python+pycuda (and other libraries) through this link: http://wiki.tiker.net/PyCuda/Installation/Linux
But when I run test program, it says:
Traceback (most recent call last):
File "test_driver.py", line 17, in <module>
import pycuda.gpuarray as gpuarray
File "/usr/local/lib/python2.7/dist-packages/pycuda-2014.1-py2.7-linux-x86_64.egg/pycuda/gpuarray.py", line 3, in <module>
import pycuda.elementwise as elementwise
File "/usr/local/lib/python2.7/dist-packages/pycuda-2014.1-py2.7-linux-x86_64.egg/pycuda/elementwise.py", line 34, in <module>
from pytools import memoize_method
File "/usr/local/lib/python2.7/dist-packages/pytools-2014.3.5-py2.7.egg/pytools/__init__.py", line 5, in <module>
from six.moves import range, zip, intern, input
ImportError: cannot import name intern
six is installed. I don't know what should I do!
I was seeing this exact same issue on Ubuntu 14.04 but didn't want to override Ubuntu's version of six due to lots of finicky dependency issues. I thought it was odd that the pytools version in the error message (2014.3.5) didn't match the version from the Ubuntu apt repo (2013.5.7).
It turns out that I had previously tried to install pycuda from source by checking out the git repository. I had also previously installed pip. Since pytools is listed as a requirement in pycuda's setup.py, pip installed its version of pytools (2014.3.5). And that's where the incompatibility between six and pytools originates.
To solve the issue, I uninstalled both pycuda and pytools using pip remove pycuda pytools and then installed pycuda using apt-get which then auto-installed the compatible version of pytools.
Just wanted to post this as an alternative solution in case anyone else prefers to keep the default Ubuntu version of six.
On a OSX system I manage to solve the problem by upgrading the six package via pip. Namely
$ pip install six --upgrade
I had the same error on Ubuntu 14.04 but neither of the tips above worked. This page recommends editing the file causing the error directly. So I edited /usr/local/lib/python2.7/dist-packages/pytools/__init__.py and changed the line:
from six.moves import range, zip, intern, input
Into
try:
from six.moves import range, zip, intern, input
except ImportError:
from six.moves import range, zip, input
Not nice editing included library files, but it got around the error.
Am new to progamming and Python, i keep getting the error below when i run my program. Someone advised i should use pip to solve it. But cant pip get installed using the cmd. Though i suceeded using Powershell but still cant make it work. How do i solve this, any tips will go along way. Thanks
Traceback (most recent call last):
File "<pyshell#0>", line 1, in <module>
from satmc import satmc
File "C:\Python27\starb_models_grid1\satmc.py", line 3, in <module>
import matplotlib.pyplot as plt
File "C:\Python27\lib\site-packages\matplotlib\__init__.py", line 110, in <module>
raise ImportError("matplotlib requires dateutil")
ImportError: matplotlib requires dateutil
Am using version 2.7.3
You need to install various packages to get numpy working correctly.
libsvm-3.17.win32-py2.7
pyparsing-2.0.1.win32-py2.7
python-dateutil-2.2.win32-py2.7
pytz-2013.9.win32-py2.7
six-1.5.2.win32-py2.7
scipy-0.13.3.win32-py2.7
numpy-MKL-1.8.0.win32-py2.7
Matplotlib
Download all the binaries from this link and install and then you will have a working numpy installation.
You're probably looking for:
pip install python-dateutil
The error states that you are missing Dateutil.
This could mean that you did not install Dateutil with pip install python-dateutil or your way to install Python modules. But if you did your pip install and still get the same error it could also mean that DateUtil is installed in a folder where your Python distribution does not expect it.
In my case (Mac OSX) I installed pip with Brew http://brew.sh/. That pip will install the modules in: /usr/local/lib/python2.7/site-packages/
However the standard Mac OSX Python distribution will look for modules in /Library/Python/2.7/site-packages/
Your folders could be different of course, but you can check where Python looks for modules with:
import sys
print sys.path
Check this and see if one of the directories in the list contains the directory where dateutil is installed.
If not add a PYTHONPATH environment variable: Like: PYTHONPATH=/usr/local/lib/python2.7/site-packages/ or the folder where your dateutil is installed
I did pip install python-dateutil, but kept getting the same error. I got past that error by installing from http://www.lfd.uci.edu/~gohlke/pythonlibs/#python-dateutil
Look for the Matplotlib section and install the dependencies you don't have.
Matplotlib is a 2D plotting library.
Requires numpy, dateutil, pytz, pyparsing, six, setuptools, and optionally pillow, pycairo, tornado, wxpython, pyside, pyqt4, ghostscript, miktex, ffmpeg, mencoder, avconv, or imagemagick.