scrubadub
Remove personally identifiable information from free text. Sometimes we have additional metadata about the people we wish to anonymize. Other times we don't. This package makes it easy to seamlessly scrub personal information from free text, without compromising the privacy of the people we are trying to protect.
scrubadub currently supports removing:
- Names
- Email addresses
- Addresses/Postal codes (US, GB, CA)
- Credit card numbers
- Dates of birth
- URLs
- Phone numbers
- Username and password combinations
- Skype/twitter usernames
- Social security numbers (US and GB national insurance numbers)
- Tax numbers (GB)
- Driving licence numbers (GB)
Quick start
Getting started with scrubadub is as easy as pip install scrubadub and
incorporating it into your python scripts like this:
>>> import scrubadub
# My cat may be more tech-savvy than most, but he doesn't want other people to know it.
>>> text = "My cat can be contacted on [email protected], or 1800 555-5555"
# Replaces the phone number and email addresse with anonymous IDs.
>>> scrubadub.clean(text)
'My cat can be contacted on {{EMAIL}}, or {{PHONE}}'There are many ways to tailor the behavior of scrubadub using
different Detectors and PostProcessors.
Scrubadub is highly configurable and supports localisation for different languages and regions.
Installation
To install scrubadub using pip, simply type:
pip install scrubadub
There are several other packages that can optionally be installed to enable extra detectors. These scrubadub_address, scrubadub_spacy and scrubadub_stanford, see the relevant documentation (address detector documentation and name detector documentation) for more info on these as they require additional dependencies. This package requires at least python 3.6. For python 2.7 or 3.5 support use v1.2.2 which is the last version with support for these versions.
New maintainers
LeapBeyond are excited to be supporting scrubadub with ongoing maintenance and development. Thanks to all of the contributors who made this package a success, but especially @deanmalmgren, IDEO and Datascope.