liberator 0.1.1

static code extractor for Python

Github	https://gitlab.kitware.com/python/liberator
Pypi	https://pypi.org/project/liberator
ReadTheDocs	https://liberator.readthedocs.io/en/latest/

Liberator is a Python library that “liberates” (i.e. statically extracts) class / function source code from an existing python library into a single standalone module.

It works by statically parsing the code for the class / function definition and then recursively parsing and extracting all missing dependencies.

Here is an example. Say that you want to use the find_exe function from ubelt, but you don’t want to depend on ubelt itself, you can extract the minimal code necessary to define the find_exe function in a new standalone module via:

# Create a Liberator instance to manage construction of
# the new standalone code.
from liberator.core import Liberator
lib = Liberator()

# Pass the liberator object the function you want to expose
# You can add more than one function this way.
import ubelt as ub
lib.add_dynamic(ub.find_exe)

# Tell liberator about the library that you dont want to depend on,
# any (statically findable) external references to that library will be
# vendored into the new liberated code.
lib.expand(['ubelt'])

# Finally access the "current source code" which will
# have the self-contained version of the "liberated" utility.
print(lib.current_sourcecode())

Limitations

Liberator has several limitations. It only works if the code is expressed statically (i.e. no decorators that register methods in a list to be injected into something later). It does not do any work to evaluate anything, it just looks at function definitions and pulls them in verbatim. It also does no work to mangle to deconflict different functions with the same names (although it could and that is an extension I’d lke add). Lastly, it can only “expand” packages imported at the top level of the code. Nested imports are not “liberated”. This is another extension that could be implemented, but it currently isn’t.