Per-field truncation / summarization for large nested log fields

[aress31]

Summary

It would be useful if structlog provided a built-in processor for per-field truncation / summarization of large values, especially nested dicts and lists.

Problem

Some fields in structured logs are much larger than others, for example:

• request / response payloads
• nested kwargs
• prompt text
• model outputs

These can overwhelm otherwise readable logs. Global string truncation is not enough; what is often needed is field-aware summarization.

Desired behavior

Something like a built-in processor where selected fields can be limited independently, for example:

• truncate long strings for prompt
• limit list length for messages
• limit depth / keys for nested dicts like payload

Example

log.debug(
    "completion.request",
    prompt=very_large_prompt,
    messages=large_messages,
    extra_body=large_nested_dict,
)

Possible API shape:

structlog.processors.truncate_fields(
    {
        "prompt": {"max_length": 300},
        "messages": {"max_items": 2, "max_string_length": 200},
        "extra_body": {"max_depth": 1, "max_keys": 10},
    }
)

Why

This seems like a common structured-logging need: preserve useful context while keeping summary/debug logs readable without writing a custom processor every time.

Would maintainers be open to something like this?