Maintenance: improve the Data Masking utility #8780

Why is this needed?

I spent some time going through the Data Masking utility, mostly erase(), and found a few edge cases where the result isn't what the docs say, plus some rough spots in error handling and docs. I'd rather fix them together than open ten small issues.

Which area does this relate to?

Other

Solution

Rough list, details in the PRs:

• fields and masking_rules passed together
• Pydantic models with SecretStr and other non-JSON field types
• JSON string input with masking_rules
• prepare_data with repeated references and __slots__ objects (Bug: data masking prepare_data leaves dataclasses and Pydantic models unconverted #8672)
• masking_rules pointing at a nested dict or list
• KMS errors wrapped in our own exceptions
• small cleanups: warning dedup in lenient mode, regex cache size, dead code in base.py
• docs for the behaviours above

#8636, #8637 and #8672 fold into this.

Acknowledgment

• This request meets Powertools for AWS Lambda (Python) Tenets
• Should this be considered in other Powertools for AWS Lambda languages? i.e. Java, TypeScript, and .NET