Search Python code for algorithmic features
Paroxython is a set of command line tools which tag and filter by algorithmic features your collection of Python programming exercises.
You are a teacher, in charge of an introductory programming course in an educational institution. Over the years, you have accumulated many—far too many—programs and code snippets that may be of interest to your students.
Or, as a seasoned developer, you would like to share your knowledge by helping a loved one learn how to code. A cursory search for pedagogical material yields an overwhelming amount of websites and repositories stuffed with Python programs of various levels (e.g., 1, 2, 3, 4, 5, 6, and a lot more from Awesome Python in Education).
In any case, the Python source codes you have gathered are typically numerous (hundreds or even thousands), reasonably sized (anything below 100 lines of code), and educational in nature (e.g., snippets, examples, quizzes, exercise solutions, classic algorithms). The programming concepts you plan to teach remain relatively low level (e.g. assignments, nested loops, accumulation patterns, tail recursive functions, etc.).
If all that sounds familiar, keep reading me.
Paroxython aims to help you select, from your collection, the one program that best suits your needs. For instance, it will gladly answer the following questions:
- How can this concept be illustrated?
- What problems use the same algorithmic and data structures as this one?
- What homework assignment should I give my students so they can practice the content of the last lesson?
Moreover, since Paroxython knows what your class knows, it can recommend the right program at the right time:
- What would make a good review exercise?
- Which exercises can I give on this exam?
- What is the current learning cost of this example?
In the long run, Paroxython may guide you and somehow make you rethink your course outline:
- What are the prerequisites for the concept of assignment?
- Do I have enough material to introduce subroutines before I even talk about conditionals and loops?
- Among the loops, which must come first: the most powerful (
while), or the most useful (
- How to logically structure this bunch of usual iterative patterns?
- What are the basics, exactly?
All issues on which the author changed his mind since he started to work on this project!
In an ideal world, Paroxython could even put an end to the deadliest religious wars, with rational, data-driven arguments:
- Father, is it a sin to exit early?
- Should a real byte use a mask?
How it works
Paroxython starts from a given folder of programs. Its contents is parsed, and all features that meet the provided specifications are labelled and associated with their spans (e.g.,
"assignment_lhs_identifier:a": 4, 6, 18 or
"loop_with_late_exit:while": 3-7, 20-29).
These labels constitute only scattered knowledge. The next step is to map them onto a taxonomy designed with basic hierarchical constraints in mind (e.g., the fact that the introduction of the concept of early exit must come after that of loop, which itself requires that of control flow, is expressed by the taxon
Extract of the taxonomy generated from The Algorithms - Python.
Click to jump to its full dynamic version in the user manual.
Everything is then persisted in a tag database, which can later be filtered through a pipeline of commands, for instance:
- include only the programs which feature a recursive function;
- exclude this or that program you want to set aside for the exam;
- “impart” all programs studied so far, i.e, consider that all the notions they implement are acquired.
The result is a list of program recommendations ordered by increasing learning cost.
Suppose that the
programs directory contains these simple programs.
First, build this tag database:
> paroxython collect programs Labelling 21 programs. Mapping taxonomy on 21 programs. Writing programs_db.json.
Then, filter it through this pipeline:
> paroxython recommend programs Processing 5 commands on 21 programs. 19 programs remaining after operation 1 (impart). 18 programs remaining after operation 2 (exclude). 12 programs remaining after operation 3 (exclude). 10 programs remaining after operation 4 (include). 10 programs remaining after operation 5 (hide). Dumped: programs_recommendations.md.
Et voilà, your recommendation report!
Installation and test-drive
Much to no one's surprise:
pip install paroxython
The following command should print a help message and exit:
IPython magic command
If you use Jupyter notebook/lab, you've also just installed a so-called magic command. Load it like this:
This should print
"paroxython 0.6.3 loaded.". Run it on a cell of Python code:
%%paroxython # Lines def fibonacci(n): # 2 result =  # 3 (a, b) = (0, 1) # 4 while a < n: # 5 result.append(a) # 6 (a, b) = (b, a + b) # 7 return result # 8
||4, 4, 7, 7|
||2-8, 2-8, 2-8, 2-8|
As you can see, in this program, Paroxython identifies among others:
- the use of the procedural paradigm (
- an impure function (
flow/loop/while) with a late exit (
- a little bit of voodoo on lists (
- a simple tuple assignment (
var/assignment/explicit/parallel). Note that we distinguish between explicit (with
=) and implicit (parameters and iteration variables) assignments;
- a “sliding” tuple assignment (
var/assignment/explicit/parallel/slide). If the denomination is unique to us, the pattern itself occurs in a number of programs: implementations of C-finite sequences with C greater than 1, Greatest Common Divisor, Quicksort, etc.
- four local variables (
- an estimation of the variety of concepts involved (
meta/count/variety/***), depending on the number of lines, features and distinct features.
The magic command
%%paroxython (corresponding to the subcommand
tag) only scratches the surface of the system. As shown before, to estimate the learning cost of the features and get actionable recommendations, you will need first to construct the tag database with
collect, and then call
recommend on a pipeline of yours.
Although this is still a work-in-progress, Paroxython should already be fairly well documented:
- User manual:
- Developer manual:
- Module reference.
- Feature specifications: a document mixing prose, tests, regular expressions and SQL queries to describe which algorithmic features are recognized and how.
- User types: all objects of interest are precisely typed and checked by mypy.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size paroxython-0.6.3-py3-none-any.whl (120.1 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size paroxython-0.6.3.tar.gz (114.7 kB)||File type Source||Python version None||Upload date||Hashes View|
Hashes for paroxython-0.6.3-py3-none-any.whl