The module_dependencies Python module allows you to gather the dependencies of specific modules in source code. It has been split into two main sections: Module and Source.
The former, Module, supports functionality for mapping a module name to the usage of that module within open source repositories.
This is very useful when we are interested in determining which sections of a Python module is most frequently used. For example:
from module_dependencies import Module
from pprint import pprint
# Attempt to find 1000 imports of the "nltk" module
# in both Python files and Jupyter Notebooks each
module = Module("nltk", count="1000")
pprint(module.usage()[:15])
module.plot()This program outputs:
[2022-01-03 14:14:39,127] [module_dependencies.module.session] [INFO ] - Fetching Python source code containing imports of `nltk`...
[2022-01-03 14:14:42,824] [module_dependencies.module.session] [INFO ] - Fetched Python source code containing imports of `nltk` (status code 200)
[2022-01-03 14:14:42,825] [module_dependencies.module.session] [INFO ] - Parsing 6,830,859 bytes of Python source code as JSON...
[2022-01-03 14:14:42,865] [module_dependencies.module.session] [INFO ] - Parsed 6,830,859 bytes of Python source code as JSON...
[2022-01-03 14:14:42,866] [module_dependencies.module.session] [INFO ] - Extracting dependencies of 725 files of Python source code...
Parsing Files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 725/725 [00:02<00:00, 258.48files/s]
[2022-01-03 14:14:45,702] [module_dependencies.module.session] [INFO ] - Extracted dependencies of 725 files of Python source code.
[2022-01-03 14:14:45,703] [module_dependencies.module.session] [INFO ] - Fetching Jupyter Notebook source code containing imports of `nltk`...
[2022-01-03 14:14:48,726] [module_dependencies.module.session] [INFO ] - Fetched Jupyter Notebook source code containing imports of `nltk` (status code 200)
[2022-01-03 14:14:48,726] [module_dependencies.module.session] [INFO ] - Parsing 25,713,281 bytes of Jupyter Notebook source code as JSON...
[2022-01-03 14:14:48,886] [module_dependencies.module.session] [INFO ] - Parsed 25,713,281 bytes of Jupyter Notebook source code as JSON...
[2022-01-03 14:14:48,888] [module_dependencies.module.session] [INFO ] - Extracting dependencies of 495 files of Jupyter Notebook source code...
Parsing Files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 495/495 [00:02<00:00, 167.09files/s]
[2022-01-03 14:14:51,851] [module_dependencies.module.session] [INFO ] - Extracted dependencies of 495 files of Jupyter Notebook source code.
[('nltk.tokenize.word_tokenize', 327),
('nltk.download', 298),
('nltk.corpus.stopwords.words', 257),
('nltk.tokenize.sent_tokenize', 126),
('nltk.stem.porter.PorterStemmer', 115),
('nltk.stem.wordnet.WordNetLemmatizer', 99),
('nltk.tag.pos_tag', 75),
('nltk.stem.snowball.SnowballStemmer', 48),
('nltk.data.path.append', 42),
('nltk.probability.FreqDist', 42),
('nltk.tokenize.RegexpTokenizer', 42),
('nltk.tokenize.TweetTokenizer', 35),
('nltk.corpus.wordnet.synsets', 33),
('nltk.data.load', 32),
('nltk.translate.bleu_score.corpus_bleu', 29)]And then opens an interactive version of the following plot:
(Note that the true plot is interactive, but this copy for GitHub is just a png)
With the methods provided in the Module class, it becomes elementary to see which sections of code are frequently used, allowing you to prioritise these sections over rarely used sections.
module_dependencies also provides Source, wich implements functionality for mapping Python source code to the dependencies and imports within that file. For example:
from module_dependencies import Source
from pprint import pprint
# This creates a Source instance for this file itself
src = Source.from_file(__file__)
pprint(src.dependencies())
pprint(src.imports())This program outputs:
['module_dependencies.Source.from_file', 'pprint.pprint']
['module_dependencies', 'pprint']More detailed documentation, including examples and an API Reference, can be found in the online documentation. I also wrote a paper about module_dependencies.
module_dependencies can be installed directly via pip. It is recommended to set up a virtualenvironment before installation, although this is not strictly a requirement.
The command to install module_dependencies is:
pip install module_dependencies
Note that module_dependencies requires Python 3.7 onwards.