Jungian.py, a theory-agnostic Python library for working with Jungian typology.

Official repo: https://github.com/kirkirz/jungian

It's available on PyPI:
pip install jungian

This was initially built for my own system (PSys), but we will gladly support other systems.
It is intended as the most comprehensive library, that anyone could use, no matter what system they support, so that we would benefit from common infrastructure.

Contributions are very welcome.

Planned structure

src/models/ : Typology model catalogue

src/lexical/ : Lexical analysis tooling

src/questionnaires/ : Questionnaires tooling

Simulations will live inside src/models/psys/

Features

Models supported: PSys, Big5, MBTI, Socionics (Models A, J)

Psys

• Psys function vectors

Big5 (aka OCEAN, CANOE and FFM)

• Big5 dimensional data structure

MBTI

• Beebe and Grant stacks
• MBTI type codes

Socionics

• Reinin dichotomies
• Model A, Model J stacks
• Function signing (legacy and modern systems)
• All 14 classical intertype relations
• Dimensionality
• Small groups

Lexical analysis

• supports custom lexicon files
• multiple weighted functions per token
• phrase handling (supports fuzzy detection) and punctuation handling

Questionnaire engine

• weighted questions with variants
• supports both dichotomy-typing and function-typing through a unified mapping system, custom flexible JSON DSL (+ config-driven scoring interpreter for trait-based models)
• CLI runner to use it interactively
• Likert scoring + normalisation of scores, reverse scoring

Quickstart for questionnaire developers

Define a questionnaire in JSON with dimensions, Likert items, reverse scoring, and weighted mappings:

{
  "name": "Cognitive Style Inventory",
  "dimensions": {
    "Ti": {"min": 0, "max": 40},
    "Ne": {"min": 0, "max": 40}
  },
  "items": [
    {"text": "I enjoy breaking systems down to their core rules", "mapping": {"Ti": 2}},
    {"text": "I find connections between seemingly unrelated ideas", "mapping": {"Ne": 2}},
    {"text": "I prefer following established procedures", "mapping": {"Ti": 1}, "reverse": true}
  ]
}

Load, score, and normalize in a few lines:

from jungian.questionnaires.quest import load_quest, score_quest, normalize_scores

data = load_quest("my_quest.json")
answers = [3, 4, 2]  # 0–4 Likert scale
raw = score_quest(data, answers)          # → {"Ti": 14, "Ne": 8}
norm = normalize_scores(raw, data)        # → {"Ti": 35.0, "Ne": 20.0}

Or run it interactively in the terminal:

python -m jungian.questionnaires.quest_cli my_quest.json

Quickstart for lexical analysis

Write a custom lexicon that maps words and phrases to weighted functions:

{
  "lexicon": {
    "logic": [["Ti", 2]],
    "data": [["Te", 2]],
    "feel": [["Fi", 1]],
    "together": [["Fe", 1]]
  },
  "phrases": {
    "think about": [["Ti", 2]],
    "what works": [["Te", 2]]
  }
}

Then analyze any text:

from jungian.lexical.guess import guess_from_file

text = "I think about how the data works in practice."
scores = guess_from_file(text, "my_lexicon.json")

print(scores)  # → {"Ti": 2, "Te": 2}

The engine handles punctuation, case-insensitivity, and fuzzy phrase matching out of the box.

Quickstart for typology enthusiasts

Check belonging to a type group:

from jungian.dichotomy import Dichotomy, E, N, T, P, I, S, F, J
from jungian.type import ENTP, INTJ, ENFP

NT = N & T

NT(ENTP) # True
NT(INTJ) # True
NT(ENFP) # False

Calculate intertype relations:

from jungian.type import LIE, ILE
from jungian.models.socionics.itr import relation

relation(LIE, ILE) # Quasi-identity

Work with stacks:

from jungian.type import INTP
from jungian.models.beebe import nemesis
print(f"Nemesis of INTP: {nemesis(INTP)}")

ROADMAP

Improving the core engine

• Make types constructed from dimensions, migrate from Dichotomy to Dimension
• Better stack tooling, including universal pos() function

Improving model support

• Add Ring, Block and Axis classes
• Tooling for other frameworks like KTS, OPS, 16P NERIS etc (there's a lot of them and implementation should be trivial once Dichotomy is polished).
• Improving the API for existing models

PSys concepts

• Adding support for Phase, Mode and Stance as standalone objects
• Implementing simulations

Side quests

• Improve testing (unit tests are necessary to ensure the correctness of derivations).
• Add more lexicon files to the repo (data is crucial here)
• Add/improve examples, improve documentation
• Lint/refactor the code and make types more rigorous (mypy), improving packaging/config