fix: handle non-numeric sort_docs_by in MetaFieldGroupingRanker

[mustafakhokhar]

Related Issues

• fixes Bug: MetaFieldGroupingRanker raises TypeError when sorting by a non-numeric metadata field #11725

Proposed Changes:

MetaFieldGroupingRanker.run() raised TypeError: '<' not supported between instances of 'float' and 'str' when sort_docs_by pointed at a non-numeric metadata field (e.g. ISO date strings) and some documents were missing that field. The sort used float("inf") as the missing-value sentinel, which cannot be compared against non-numeric values.

This resolves the sort field once and uses a type-safe (value is None, value) sort key, so documents missing the field (or whose value is None) are placed at the end of their group regardless of the field's type. Behaviour for numeric fields is unchanged.

Repro (raises on main):

from haystack.components.rankers import MetaFieldGroupingRanker
from haystack.dataclasses import Document

docs = [
    Document(content="newest", meta={"group": "42", "date": "2023-03-01"}),
    Document(content="missing date", meta={"group": "42"}),
    Document(content="oldest", meta={"group": "42", "date": "2023-01-01"}),
]
MetaFieldGroupingRanker(group_by="group", sort_docs_by="date").run(documents=docs)
# TypeError: '<' not supported between instances of 'float' and 'str'

How did you test it?

Added two unit tests in test/components/rankers/test_meta_field_grouping_ranker.py:

• sorting by a non-numeric field with a missing key
• sorting by a field that is present but set to None

All existing ranker unit tests pass. hatch run test:types (mypy) and hatch run fmt are clean, and the pre-commit hooks pass.

Notes for the reviewer

• Documents missing the sort field (or with a None value) are placed last, matching the previous numeric behaviour and the class's "documents without a group are placed at the end" convention. Sort stability is preserved.
• Out of scope: a single group mixing value types for the same key (e.g. int and str) remains inherently unsortable — that limitation is unchanged.

Checklist

• I have read the contributors guidelines and the code of conduct.
• I have updated the related issue with new insights and changes.
• I have added unit tests and updated the docstrings.
• I've used a conventional commit type for my PR title (fix:).
• I have documented my code.
• I have added a release note file, following the contributors guidelines.
• I have run pre-commit hooks and fixed any issue.

---

Implemented with the help of an AI assistant; I have reviewed all changes and run the relevant tests and quality checks.

[vercel]

@mustafakhokhar is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

All committers have signed the CLA.