lazy_import

lazy_import provides a set of functions that load modules, and related attributes, in a lazy fashion. This allows deferring of ImportErrors to actual module use-time. Likewise, actual module initialization only takes place at use-time. This is useful when using optional dependencies with heavy loading times and/or footprints, since that cost is only paid if the module is actually used.

For minimal impact to other code running in the same session lazy_import functionality is implemented without the use of import hooks.

lazy_import is compatible with Python ≥ 2.7 or ≥ 3.4.

Examples: lazy module loading

import lazy_import
np = lazy_import.lazy_module("numpy")
# np is now available in the namespace and is listed in sys.modules under
#  the 'numpy' key:
import sys
sys.modules['numpy']
# The module is present as "Lazily-loaded module numpy"

# Subsequent imports of the same module return the lazy version present
#  in sys.modules
import numpy # At this point numpy and np point to the same lazy module.
# This is true for any import of 'numpy', even if from other modules!

# Accessing attributes causes the full loading of the module ...
np.pi
# ... and the module is changed in place. np and numpy are now
#  "<module 'numpy' from '/usr/local/lib/python/site-packages/numpy/__init__.py'>"

# Lazy-importing a module that's already fully loaded returns the full
#  module instead (even if it was loaded elsewhere in the current session)
#  because there's no point in being lazy in this case:
os = lazy_import.lazy_module("os")
# "<module 'os' from '/usr/lib/python/os.py'>"

In the above code it can be seen that issuing lazy_import.lazy_module("numpy") registers the lazy module in the session-wide sys.modules registry. This means that any subsequent import of numpy in the same session, while the module is still not fully loaded, will get served a lazy version of the numpy module. This will happen also outside the code that calls lazy_module:

import lazy_import
np = lazy_import.lazy_module("numpy")
import module_that_uses_numpy # This module will get a lazy module upon
                              # 'import numpy'

Normally this is ok because the lazy module will behave pretty much as the real thing once fully-loaded. Still, it might be a good practice to document that you're lazily importing modules so-and-so, so that users are warned.

Further uses are to delay ImportErrors:

import lazy_import
# The following succeeds even when asking for a module that's not available
missing = lazy_import.lazy_module("missing_module")

missing.some_attr # This causes the full loading of the module, which now fails.
"ImportError: __main__ attempted to use a functionality that requires module
 missing_module, but it couldn't be loaded. Please install missing_module and retry."

Submodules work too:

import lazy_import
mod = lazy_import.lazy_module("some.sub.module")
# mod now points to the some.sub.module lazy module
#  equivalent to "from some.sub import module as mod"

# Alternatively the returned reference can be made to point to the
#  base module:
some = lazy_import.lazy_module("some.sub.module", level="base")

# This is equivalent to "import some.sub.module" in that only the base
#  module's name is added to the namespace. All submodules must be accessed
#  via that:
some.sub # Returns lazy module 'some.sub' without triggering full loading.
some.sub.attr # Triggers full loading of 'some' and 'some.sub'.
some.sub.module.function() # Triggers loading also of 'some.sub.module'.

Finally, if you want to mark some modules and submodules your package imports as always being lazy, it is as simple as lazily importing them at the root __init__.py level. Other files can then import all modules normally, and those that have already been loaded as lazy in __init__.py will remain so:

# in __init__.py:

import lazy_import
lazy_import.lazy_module("numpy")
lazy_import.lazy_module("scipy.stats")


# then, in any other file in the package just use the imports normally:

import requests # This one is not lazy.
import numpy # This one is lazy, as long as no other code caused its
             #  loading in the meantime.
import scipy # This one is also lazy. It was lazily loaded as part of the
             #  lazy loading of scipy.stats.
import scipy.stats # Also lazy.
import scipy.linalg # Uh-oh, we didn't lazily import the 'linalg' submodule
                    #  earlier, and importing it like this here will cause
                    #  both scipy and scipy.linalg (but not scipy.stats) to
                    #  immediately become fully loaded.

Examples: lazy callable loading

To emulate the from some.module import function syntax lazy_module provides lazy_callable. It returns a wrapper function. Only upon being called will it trigger the loading of the target module and the calling of the target callable (function, class, etc.).

import lazy_import
fn = lazy_import.lazy_callable("numpy.arange")
# 'numpy' is now in sys.modules and is 'Lazily-loaded module numpy'

fn(10)
# array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

lazy_callable is only useful when the target callable is going to be called:

import lazy_import
cl = lazy_import.lazy_callable("numpy.ndarray") # a class

obj = cl([1, 2]) # This works OK (and also triggers the loading of numpy)

class MySubclass(cl): # This fails because cl is just a wrapper,
    pass              #  not an actual class.

Installation

pip install lazy_import

Or, to include dependencies needed to run regression tests:

pip install lazy_import[test]

Tests

The lazy_module module comes with a series of tests. If you install with test dependencies (see above), just run

import lazy_import.test_lazy
lazy_import.test_lazy.run()
# This will automatically parallelize over the available number of cores

Alternatively, tests can be run from the command line:

pytest -n 4 --boxed -v --pyargs lazy_import
# (replace '4' with the number of cores in your machine, or set to 1 if
#  you'd rather test in serial)

Tests depend only on pytest and pytest-xdist, so if you didn't install them along lazy_import (as described under Installation) just run

pip install pytest pytest-xdist

Note that pytest-xdist is required even for serial testing because of its --boxed functionality.

License

lazy_import is released under GPL v3. It was based on code from the importing module from the PEAK package. The licenses for both lazy_import and the PEAK package are included in the LICENSE file. The respective license notices are reproduced here:

lazy_import — a module to allow lazy importing of python modules

Copyright (C) 2017-2018 Manuel Nuno Melo

lazy_import is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

lazy_import is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with lazy_import. If not, see <http://www.gnu.org/licenses/>.

The PEAK importing code is

Copyright (C) 1996-2004 by Phillip J. Eby and Tyler C. Sarna. All rights reserved. This software may be used under the same terms as Zope or Python. THERE ARE ABSOLUTELY NO WARRANTIES OF ANY KIND. Code quality varies between modules, from "beta" to "experimental pre-alpha". :)

Code pertaining to lazy loading from PEAK importing was included in lazy_import, modified in a number of ways. These are detailed in the CHANGELOG file of lazy_import. Changes mainly involved Python 3 compatibility, extension to allow customizable behavior, and added functionality (lazy importing of callable objects).