CerbIA
prerelease Prerelease 0.3.0 Latest
↑↓ navigate · ↵ select · esc close
CerbIA
CerbIA
prerelease Prerelease 0.3.0 Latest

cerbia-presidio

Add the optional Presidio and spaCy integration for English NLP-based PII scanning.

cerbia-presidio provides cerbia.presidio.scanners.PresidioPiiScanner, an optional English natural-language-processing scanner backed by Presidio and the en_core_web_sm spaCy model.

Install

pip install cerbia-presidio

The distribution declares its Presidio analyzer and English spaCy model dependencies. Scanner construction initializes the model, so allow for that runtime cost when choosing where to create the scanner.

Configure the scanner

By default, the scanner asks the English analyzer for every supported entity type. Set entities to limit the scan. The result uses the highest Presidio confidence as its risk score, includes detected match spans, and blocks findings by default.

scanners:
  - scanner: cerbia.presidio.scanners.PresidioPiiScanner
    init_args:
      entities: ["EMAIL_ADDRESS", "PHONE_NUMBER"]

The class is also available from cerbia.presidio.scanners:

from cerbia.presidio.scanners import PresidioPiiScanner

Use the core PiiScanner when an NLP runtime is not appropriate. For product-wide setup, see Configuration and the quickstart. The Presidio scanner reference describes the scanner alongside the other pipeline components.