24

Assuming my_function() is located in my_apps.views I would like to import my_function dynamically without using something like exec or eval.

Is there anyway to accomplish this. I'm looking to do something similar to:

my_function = import_func("my_apps.views.my_function")

my_function()
   ... code is executed
Anshuman Dikhit
  • 459
  • 2
  • 10
Pyther
  • 241
  • 1
  • 2
  • 3

6 Answers6

35

you want 

my_function = getattr(__import__('my_apps.views'), 'my_function')

If you happen to know the name of the function at compile time, you can shorten this to

my_function = __import__('my_apps.views').my_function

This will load my_apps.views and then assign its my_function attribute to the local my_function.

If you are sure that you only want one function, than this is acceptable. If you want more than one attribute, you can do:

views = __import__('my_apps.views')
my_function = getattr(views, 'my_function')
my_other_function = getattr(views, 'my_other_function')
my_attribute = getattr(views, 'my_attribute')

as it is more readable and saves you some calls to __import__. again, if you know the names, the code can be shortened as above.

You could also do this with tools from the imp module but it's more complicated.

Nathaniel Ford
  • 20,545
  • 20
  • 91
  • 102
aaronasterling
  • 68,820
  • 20
  • 127
  • 125
  • 3
    There will be problems (AttributeError) with the __import__ statements above. Import returns the first module (my_apps) unless an import list is specified. You need something like: `my_function = __import__('my_apps.views', globals(), locals(), ['my_function']).my_function` See [python doc about __import__](https://docs.python.org/2/library/functions.html#__import__) – marco Mar 10 '16 at 00:53
  • For me this attempts to import the function from `my_apps` instead of `my_apps.views`. – oarfish Nov 17 '17 at 10:18
  • @marco Actually __import__('my_apps.views', fromlist=['my_function']).my_function is enough. – oszkar Sep 06 '22 at 14:54
9

Note that Python 2.7 added the importlib module, convenience wrappers for __import__() and a backport of 3.1 feature.

This module is a minor subset of what is available in the more full-featured package of the same name from Python 3.1 that provides a complete implementation of import. What is here has been provided to help ease in transitioning from 2.7 to 3.1.

importlib.import_module(name, package=None)

Import a module. The name argument specifies what module to import in absolute or relative terms (e.g. either pkg.mod or ..mod). If the name is specified in relative terms, then the package argument must be specified to the package which is to act as the anchor for resolving the package name (e.g. import_module('..mod', 'pkg.subpkg') will import pkg.mod). The specified module will be inserted into sys.modules and returned.

Community
  • 1
  • 1
gimel
  • 83,368
  • 10
  • 76
  • 104
  • The subset of importlib available in Python 2.7 is also available on PyPI and is backported there to work with Python 2.3 and later. – Brett Cannon Sep 20 '10 at 22:33
3
def import_by_string(full_name):
    module_name, unit_name = full_name.rsplit('.', 1)
    return getattr(__import__(module_name, fromlist=['']), unit_name)


exists = import_by_string("os.path.exists")
Archibald
  • 1,305
  • 1
  • 14
  • 19
  • Since version 2.4 (from 2004), Python has had the function `rsplit()`, it is much more efficient to do `module_name, unit_name = full_name.rsplit('.', 1)`, than what you are doing. – Anthon Mar 04 '17 at 08:35
  • Thanks. Did not think about it. – Archibald Mar 04 '17 at 18:12
1

I just wrote this code and seems what a lot of people need, so even if later i show it

def my_import(module_name,func_names = [],cache = False):
    if module_name in globals() and cache:
        return True
    try: 
        m = __import__(module_name, globals(), locals(), func_names, -1)
        if func_names:
            for func_name in func_names:
                globals()[func_name] = getattr(m,func_name)
        else:
            globals()[module_name] = m
        return True
    except ImportError:
        return False
def my_imports(modules):
    for module in modules:
        if type(module) is tuple:
            name = module[0]
            funcs = module[1]
        else:
            name = module
            funcs = []
        if not my_import(name, funcs):
             return module
    return ''

def checkPluginsImports(plugin,modules):
    c = my_imports(modules)
    if c:
        print plugin +" has errors!: module '"+c+"' not found"

# example: file test.py with "x" function
def d():
    checkPluginsImports('demoPlugin',[('test',['x'])])

d()
x()
user2054758
  • 321
  • 3
  • 18
0

Use the standard library pkg_resources

from pkg_resources import EntryPoint
my_function  = EntryPoint.parse("my_function=my_apps.views:my_function").load(require=False)
Daniel Braun
  • 2,452
  • 27
  • 25
0

We have four cases separated by the fact whether the module and/or the function fixed or not:

  1. module name is a fixed string, function name is a fixed string: 
    my_function = __import__('my_apps.views', fromlist=['my_function'].my_function
    (altough in this case it is much more simple to use from my_app.views import my_function)
  2. module name is a fixed string, function name is variable: 
    function_name = ...
.
.
.
my_function = getattr(__import__('my_apps.views', fromlist=[function_name]),
                      function_name)
  3. module name is variable, function name is fixed string: 
    module_name = ...
.
.
.
my_function = __import__(module_name, fromlist=['my_function']).my_function
  4. module name is variable, function name is variable: 
    module_name = ...
.
.
.
function_name = ...
.
.
.
my_function = getattr(__import__(module_name, fromlist=[function_name]),
                      function_name)

Note: For an empty list (that's the default value) as __import__ keyword argument fromlist not the module, but the package root is returned. For all non-empty lists the actual module returned.

Sources and further information:

oszkar
  • 865
  • 5
  • 21