isomorphism

Reversible anonymisation of Swiss personal data for Python. It replaces AHV numbers, IBANs, UID numbers, phone numbers and email addresses with placeholders before a text goes to a language model, and puts the real values back into the answer.

• Local only. No network calls, no telemetry, no dependencies, nothing downloaded at runtime.
• Reversible. anonymize(text) returns the text with placeholders and a mapping, restore(text, mapping) undoes it.
• Same value, same placeholder. The model can follow that [AHV_1] appears three times.
• Checksums. AHV (EAN-13), IBAN (mod 97) and UID (mod 11) are verified. Numbers that look right but fail the check are reported anyway, with lower confidence.
• Audit-friendly. Every hit comes back as kind, position, confidence and recogniser, without the value.

Python 3.12 or newer. MIT license.

pip install isomorphism

from isomorphism import anonymize, restore

text = "Anna Muster, AHV 756.1234.5678.97, IBAN CH93 0076 2011 6238 5295 7, Tel. 071 123 45 67"
safe, mapping = anonymize(text, lang="de")
# 'Anna Muster, AHV [AHV_1], IBAN [IBAN_1], Tel. [PHONE_1]'

answer = call_your_model(safe)  # the model only sees placeholders
print(restore(answer, mapping))  # real values are back

The mapping holds the real values. This package never logs or stores it; keeping it (in memory, per request) is up to you. repr(mapping) shows only the number of placeholders.

What it finds

detect(text) returns the hits without changing the text. Passing keep=["info@gemeinde.example"] leaves exact values untouched (case-insensitive), for example the address of an authority.

Measured on the test cases

tests/cases.py holds invented Swiss texts (payslip extract, client email, receipt, tax query, French and Italian letters, written-out variants, numbers with wrong check digits, and a text full of look-alikes that must not match). The numbers below come from python -m tests.measure, and a test fails if they are out of date or if a listed value slips through.

This is a small invented set, not a benchmark on real documents. Read it as a regression guard, and test with your own (invented) samples before you rely on it.

Not in this version

Personal names, postal addresses, company names, dates of birth, account and card numbers, and restoring a streamed answer arrive in the next version. Until then names and addresses stay in the text. lang ("de", "fr", "it", "en") is accepted and checked already but has no effect yet. Foreign IBANs and the old 11-digit AHV format are not recognised.

No tool finds all personal data. Treat the result as one protective layer, not as a guarantee of anonymity.

Development

uv venv && uv pip install -e . pytest ruff mypy
ruff check . && ruff format --check . && mypy && pytest

Docs: https://packages.sweber.dev/isomorphism/docs