I have an python application that uses the Python Imaging Library (PIL) and is packaged using Py2app. Numpy dylibs are found in the app contents directory:
demo.
py2app
implements special handling of certain packages via a mechanism called "recipes". It comes with built-in recipes for certain packages including numpy
, which is why numpy is excluded from the .zip
file. (Here's the built-in numpy recipe.)
In addition to the built-in recipes, you can define your own recipes for py2app to use. Just define a class that has a check
method, and monkey-patch it as an attribute onto py2app.recipes
:
# In setup.py
class PIL_recipe:
def check(self, cmd, mf):
m = mf.findNode("PIL")
if m is None or m.filename is None:
return None
# Exclude from site-packages.zip
return {"packages": ["PIL"]}
py2app.recipes.PIL = PIL_recipe()
...
setup(...)
If you need to do this with several libraries, you can generalize this trick so that it doesn't hard-code the name of the package:
class ExcludeFromZip_Recipe(object):
def __init__(self, module):
self.module = module
def check(self, cmd, mf):
m = mf.findNode(self.module)
if m is None:
return None
# Don't put the module in the site-packages.zip file
return {"packages": [self.module]}
for module in ['PIL', 'skimage', 'sklearn', 'jsonschema']:
setattr( py2app.recipes, module, ExcludeFromZip_Recipe(module) )
Disclaimer: This is how I've solved this problem in the past. But I'm not sure why your zip file is named python3.zip
instead of site-packages.zip
.