A simple API usage analyser. The goal of this project is to provide a tool that is very simple to read and understand, in the hope that this increases the chances of people running it on their private code bases.
To analyse source code you need the kamal.py script and its only
dependency: pip install jedi. Run python kamal.py --module <module> <path_to_code>
in an environment that has the <module> you are interested in installed.
To analyse all the Python files in a folder run:
python kamal.py --module <module_name> path/to/python/files
This will produce a statistics.csv containing one line for each call to a function
or instantiation of a class in <module_name>.
A useful tool for exploring the output of kamal is datasette.
- Create a sqlite database from the CSV with
sqlite-utils insert --csv statistics.db sklearn statistics.csv - Launch datasette:
datasette serve statistics.db
To combine the results of multiple kamal.py runs can be combined into one database
by running sqlite-utils insert --csv statistics.db sklearn <CSV> for each CSV. Each
CSV is appended to the database table.