Welcome to depfinder documentation!¶

depfinder¶

Installation¶

depfinder is on pypi. It is tested against Python 2.7 and 3.6-3.8.

pip install depfinder

It is available via conda.

conda install -c conda-forge depfinder

It is also via github.

git clone git@github.com:ericdill/depfinder
cd depfinder
python setup.py install

Using depfinder¶

$ depfinder -h
usage: depfinder [-h] [-y] [-V] [--no-remap] [-v] [-q] [-k KEY] [--conda]
                [--pdb]
                file_or_directory

Tool for inspecting the dependencies of your python project.

positional arguments:
    file_or_directory  Valid options are a single python file, a single jupyter
                        (ipython) notebook or a directory of files that include
                        python files

optional arguments:
    -h, --help         show this help message and exit
    -y, --yaml         Output in syntactically valid yaml when true. Defaults to
                        False
    -V, --version      Print out the version of depfinder and exit
    --no-remap         Do not remap the names of the imported libraries to their
                        proper conda name
    -v, --verbose      Enable debug level logging info from depfinder
    -q, --quiet        Turn off all logging from depfinder
    -k KEY, --key KEY  Select some or all of the output keys. Valid options are
                        'required', 'optional', 'builtin', 'relative', 'all'.
                        Defaults to 'all'
    --conda            Format output so it can be passed as an argument to conda
                        install or conda create
    --pdb              Enable PDB debugging on exception

Ok, great. That’s the help output. Not super helpful. What does the output of depfinder look like when we run it on the source code for depfinder?

$ depfinder depfinder
{'builtin': ['__future__',
                'argparse',
                'ast',
                'collections',
                'copy',
                'errno',
                'json',
                'logging',
                'os',
                'pprint',
                're',
                'subprocess',
                'sys'],
    'relative': ['_version', 'main'],
    'required': ['pyyaml', 'stdlib-list']}

So, what do these things mean? Well builtin are modules that are built in to the standard library. required are modules that are not from the standard library or from within the depfinder package and relative are modules that are imported from one module to another within the depfinder source code.

Also see this notebook

Description¶

Find all the unique imports in your library, automatically, because who likes do it by hand? depfinder uses the ast (Abstract Syntax Tree) module (and more ast docs) to find all :pyast.Try{.interpreted-text role=”class”} and :pyast.ImportFrom{.interpreted-text role=”class”} nodes. These :pyast.Import{.interpreted-text role=”class”} and :pyast.ImportFrom{.interpreted-text role=”class”} nodes are then grouped according to the following categories, in order of decreasing precedence:

relative : The import is a relative import from within the same library
builtin : The import is built into the standard library, as determined by scraping the official python docs for the builtins with stdlib-list
questionable : The import occurs inside any combination of
- ast.Try (ast.TryExcept on py27)
- ast.FunctionDef or ast.AsyncFunctionDef
- ast.If, ast.While, ast.For, or ast.AsyncFor.
- ast.match_case.
The module may be importable without these imports, but the it will likely not have full functionality.
required : The import occurs at the top level of the module and will get executed when the module is imported. These imports must be accounted for in an environment, or the module will not be importable.

Testing¶

It has dependencies on, stdlib-list and pyyaml. I use stdlib-list to get the list of libraries built in to the standard library. These requirements can be installed via pip :

pip install -r requirements.txt

Also install the test-requiements :

pip install -r test-requirements.txt

Then you can run the tests from the root of the git repository :

coverage run run_tests.py

Releasing¶

manual:

create an annotated tag and push it to github. git tag -a TAG and then git push –tags`
git checkout TAG
python -m build --sdist --wheel . --outdir dist
twine check dist/*
twine upload dist/* --verbose

API¶

depfinder.main.get_imported_libs(code, filename='', custom_namespaces=None)[source]¶

Given a code snippet, return a list of the imported libraries

Parameters:

codestr: The code to parse and look for imports

Returns:

ImportCatcher: The ImportCatcher is the object in depfinder that contains all the information regarding which imports were found where. You will most likely be interested in calling the describe() function on this return value.

Examples

>>> depfinder.get_imported_libs('from foo import bar')
{'required': {'foo'}, 'questionable': set()}
>>> with open('depfinder.py') as f:
        code = f.read()
        imports = depfinder.get_imported_libs(code)
        print(imports.describe())
{'builtin': {'__future__', 'json', 'ast', 'os', 'sys', 'collections'},
 'required': {'stdlib_list'}}

depfinder.main.iterate_over_library(path_to_source_code, custom_namespaces=None)[source]¶

Helper function to recurse into a library and find imports in .py files.

This allows the user to apply filters on the user-side to exclude imports based on their file names. conda-skeletor makes heavy use of this function

Parameters:

path_to_source_codestr

Yields:

catcherstuple: Yields tuples of (module_name, full_path_to_module, ImportCatcher)

depfinder.main.simple_import_search(path_to_source_code, remap=True, ignore=None, custom_namespaces=None)[source]¶

Return all imported modules in all .py files in path_to_source_code

Parameters:

path_to_source_codestr
remapbool, optional: Normalize the import names to be synonymous with their conda/pip names
ignorelist, optional: String pattern which if matched causes the file to not be inspected
custom_namespaceslist of str or None: If not None, then resulting package outputs will list everying under these namespaces (e.g., for packages foo.bar and foo.baz, the outputs are foo.bar and foo.baz instead of foo if custom_namespaces=[“foo”]).

Returns:

dict: The list of all imported modules, sorted according to the keys listed in the docstring of depfinder.ImportCatcher.describe()

Examples

>>> depfinder.simple_import_search('/path/to/depfinder/source')
{'builtin': ['__future__',
             'ast',
             'collections',
             'json',
             'os',
             'shlex',
             'sys',
             'tempfile'],
 'required': ['depfinder',
              'nbformat',
              'pytest',
              'setuptools',
              'sphinx_rtd_theme',
              'stdlib_list',
              'test_with_code']}

IPython/Jupyter Notebook support¶

depfinder has support for v4 Jupyter notebooks.

depfinder.main.notebook_path_to_dependencies(path_to_notebook, remap=True, custom_namespaces=None)[source]¶

Helper function that turns a jupyter notebook into a list of dependencies

Parameters:

path_to_notebookstr
remapbool, optional: Normalize the import names to be synonymous with their conda/pip names
custom_namespaceslist of str ot None, optional: If not None, then resulting package outputs will list everying under these namespaces (e.g., for packages foo.bar and foo.baz, the outputs are foo.bar and foo.baz instead of foo if custom_namespaces=[“foo”]).

Returns:

dict

Dict of dependencies keyed on

‘builtin’ - libraries built in to python
‘required’ - libraries that are found at the top level of your modules
‘questionable’ - libraries that are found inside try/except blocks
‘relative’ - libraries that are relative imports

Examples

>>> depfinder.notebook_path_to_dependencies('depfinder_usage.ipynb')
{'builtin': ['os', 'pprint'], 'required': ['depfinder']}

Welcome to depfinder documentation!¶

depfinder¶

Installation¶

Using depfinder¶

Description¶

Testing¶

Releasing¶

API¶

IPython/Jupyter Notebook support¶

Indices and tables¶