83

I have a file called foobar (without .py extension). In the same directory I have another python file that tries to import it:

import foobar

But this only works if I rename the file to foobar.py. Is it possible to import a python module that doesn't have the .py extension?

Update: the file has no extension because I also use it as a standalone script, and I don't want to type the .py extension to run it.

Update2: I will go for the symlink solution mentioned below.

compie
  • 10,135
  • 15
  • 54
  • 78

7 Answers7

62

You can use the imp.load_source function (from the imp module), to load a module dynamically from a given file-system path.

import imp
foobar = imp.load_source('foobar', '/path/to/foobar')

This SO discussion also shows some interesting options.

Rob Kwasowski
  • 2,690
  • 3
  • 13
  • 32
Eli Bendersky
  • 263,248
  • 89
  • 350
  • 412
  • Fixed. [it is more constructive to suggest an Edit, though] – Eli Bendersky May 14 '15 at 14:02
  • but again section is required as I understood from >>> foobar = imp.load_source('','credentials') [default] NameError: name 'default' is not defined – Ilja Oct 11 '15 at 10:16
  • 2
    What is the use of the first argument `'foobar'` if it is assigned in the return value? – Anmol Singh Jaggi Apr 30 '16 at 14:13
  • 2
    @AnmolSinghJaggi It sets the module `__name__` property; normally that's determined from the filename, but since you're using a non-standard filename (which might not even contain any valid python identifier at all), you have to specify the module name. The variable name in which you store a reference to the created module object is irrelevant, much as if you `import foo.bar as baz` the module referenced by the variable `baz` will still have its original `__name__`. – Ben Apr 26 '17 at 02:40
  • 3
    This has been deprecated since 3.4. Any idea how to import from a file without the .py extension in 3.4+? – exhuma Oct 06 '17 at 09:42
  • 2
    For Python 3.4+: this answer is more exact: https://stackoverflow.com/questions/2601047/import-a-python-module-without-the-py-extension/43602645#43602645 – tres.14159 Oct 29 '18 at 12:48
34

Here is a solution for Python 3.4+:

from importlib.util import spec_from_loader, module_from_spec
from importlib.machinery import SourceFileLoader 

spec = spec_from_loader("foobar", SourceFileLoader("foobar", "/path/to/foobar"))
foobar = module_from_spec(spec)
spec.loader.exec_module(foobar)

Using spec_from_loader and explicitly specifying a SourceFileLoader will force the machinery to load the file as source, without trying to figure out the type of the file from the extension. This means that you can load the file even though it is not listed in importlib.machinery.SOURCE_SUFFIXES.

If you want to keep importing the file by name after the first load, add the module to sys.modules:

sys.modules['foobar'] = foobar

You can find an implementation of this function in a utility library I maintain called haggis. haggis.load.load_module has options for adding the module to sys.modules, setting a custom name, and injecting variables into the namespace for the code to use.

Mad Physicist
  • 107,652
  • 25
  • 181
  • 264
  • 4
    That module needs a `foobar = importlib.nice_import('foobar')` helper desperately. – Ciro Santilli OurBigBook.com Sep 08 '18 at 08:08
  • @Ciro. It already has that. I don't think that `/some/arbitrary/file.weird_extension` qualifies as "nice". That being said, I've started using python code for all my configuration files once I discovered this. It's just so convenient. – Mad Physicist Sep 08 '18 at 14:04
  • What if I want to import * ? – Alex Harvey Feb 27 '20 at 04:51
  • 1
    @AlexHarvey. This gives you a module object. You can do something like `globals().update(foobar.__dict__)` or so, but I would recommend against it. – Mad Physicist Feb 27 '20 at 13:24
  • I'm trying to do this as a single executable and import it just for testing. A bit disappointed at the answers, too much speculation, too few concrete answers... is there a 'nice_import' ? Name it! – OriginalHacker Jan 31 '22 at 09:22
  • @OriginalHacker. I'm not sure where you see speculation. This is how you use existing import machinery to load an arbitrary file as though it were a python file. Not sure what you mean by "nice_import". That being said, if you're not happy with the solution being three lines, you can check out [`haggis.load.load_module`](https://haggis.readthedocs.io/en/latest/api.html#haggis.load.load_module) – Mad Physicist Jan 31 '22 at 09:47
18

Like others have mentioned, you could use imp.load_source, but it will make your code more difficult to read. I would really only recommend it if you need to import modules whose names or paths aren't known until run-time.

What is your reason for not wanting to use the .py extension? The most common case for not wanting to use the .py extension, is because the python script is also run as an executable, but you still want other modules to be able to import it. If this is the case, it might be beneficial to move functionality into a .py file with a similar name, and then use foobar as a wrapper.

  • 15
    Or instead of wrapping, just symlink foobar.py to foobar (assuming you aren't on Windows) – whaley Apr 08 '10 at 18:31
  • @whaley, yeah, that would be much cleaner. You could use a .bat for windows to accomplish the same thing. –  Apr 08 '10 at 22:45
  • I've got a neat use case - a readme file, with examples in it, which I'd like doctest to validate. I'm hoping to make a doctest markdown doc that works... – Danny Staple Nov 24 '11 at 23:05
  • And the answer is (for that use case) - use doctest.loadfile! – Danny Staple Nov 24 '11 at 23:10
  • The issue with wrappers is that if someone naively copies just the wrapper to a `bin/` directory, the program won't work when run from the path. – cjs Nov 22 '17 at 01:14
14

imp.load_source(module_name, path) should do or you can do the more verbose imp.load_module(module_name, file_handle, ...) route if you have a file handle instead

Daniel DiPaolo
  • 55,313
  • 14
  • 116
  • 115
8

importlib helper function

Here is a convenient, ready-to-use helper to replace imp, with an example, based on what was mentioned at: https://stackoverflow.com/a/43602645/895245

main.py

#!/usr/bin/env python3

import os
import importlib
import sys

def import_path(path):
    module_name = os.path.basename(path).replace('-', '_')
    spec = importlib.util.spec_from_loader(
        module_name,
        importlib.machinery.SourceFileLoader(module_name, path)
    )
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)
    sys.modules[module_name] = module
    return module

notmain = import_path('not-main')
print(notmain)
print(notmain.x)

not-main

x = 1

Run:

python3 main.py

Output:

<module 'not_main' from 'not-main'>
1

I replace - with _ because my importable Python executables without extension have hyphens. This is not mandatory, but produces better module names.

This pattern is also mentioned in the docs at: https://docs.python.org/3.7/library/importlib.html#importing-a-source-file-directly

I ended up moving to it because after updating to Python 3.7, import imp prints:

DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses

and I don't know how to turn that off, this was asked at:

Tested in Python 3.7.3.

alex
  • 3
  • 2
Ciro Santilli OurBigBook.com
  • 347,512
  • 102
  • 1,199
  • 985
2

If you install the script with package manager (deb or alike) another option would be to use setuptools:

"...there’s no easy way to have a script’s filename match local conventions on both Windows and POSIX platforms. For another, you often have to create a separate file just for the “main” script, when your actual “main” is a function in a module somewhere... setuptools fixes all of these problems by automatically generating scripts for you with the correct extension, and on Windows it will even create an .exe file..."

https://pythonhosted.org/setuptools/setuptools.html#automatic-script-creation

user2745509
  • 411
  • 4
  • 5
0

import imp has been deprecated.

The following is clean and minimal for me:

import sys
import types
import  pathlib

def importFileAs(
        modAsName: str,
        importedFilePath: typing.Union[str,  pathlib.Path],
) -> types.ModuleType:
    """ Import importedFilePath as modAsName, return imported module
by loading importedFilePath and registering modAsName in sys.modules.
importedFilePath can be any file and does not have to be a .py file. modAsName should be python valid.
Raises ImportError: If the file cannot be imported or any Exception: occuring during loading.

Refs:
Similar to: https://stackoverflow.com/questions/19009932/import-arbitrary-python-source-file-python-3-3
    but allows for other than .py files as well through importlib.machinery.SourceFileLoader.
    """
    import importlib.util
    import importlib.machinery

    # from_loader does not enforce .py but  importlib.util.spec_from_file_location() does.
    spec = importlib.util.spec_from_loader(
        modAsName,
        importlib.machinery.SourceFileLoader(modAsName, importedFilePath),
    )
    if spec is None:
        raise ImportError(f"Could not load spec for module '{modAsName}' at: {importedFilePath}")
    module = importlib.util.module_from_spec(spec)

    try:
        spec.loader.exec_module(module)
    except FileNotFoundError as e:
        raise ImportError(f"{e.strerror}: {importedFilePath}") from e

    sys.modules[modAsName] = module
    return module

And then I would use it as so:

aasMarmeeManage = importFileAs('aasMarmeeManage', '/bisos/bpip/bin/aasMarmeeManage.cs')
def g_extraParams(): aasMarmeeManage.g_extraParams()
Mohsen Banan
  • 85
  • 1
  • 5