Search, Filter, and Group Metadata

Scribd's discovery systems receive an unsorted array of document metadata. Implement a function that searches documents by text, applies dynamic filters, and groups the matching document IDs by a requested metadata field.

The search is case-insensitive and matches the document's title, author, or any value in tags. A blank or None query matches every document. Filters are supplied as a dictionary:

1. A scalar value requires exact equality.
2. A list requires a scalar document field to appear in that list, or requires at least one overlap when the document field is a list such as tags.
3. A range object such as {"min": 2015, "max": 2024} applies inclusive bounds. Either bound may be omitted.

Return a dictionary mapping each group value to a list of matching document IDs. Preserve the input order within each group and omit groups with no matches.

Formal Specification

Implement search_filter_group(documents, query, filters, group_by), where documents is a list of dictionaries, query is a string or None, filters is a dictionary, and group_by is a metadata field name. Return a dictionary whose keys are string representations of group values and whose values are lists of IDs.