Add Sparse Score Contributions

Task

Write add_sparse_score_contributions(documents, contributions). documents is an ordered list of unique document IDs. contributions is a mapping from a term name to a mapping of document ID to one finite numeric contribution. A term mapping contains only documents that received a contribution; a document may be absent from every term. The insertion order of the outer mapping is the term order used in the trace. Do not construct a dense document-by-term array.

For each document, add all supplied contributions. Retain the evidence in a trace whose document entries list {"term": term, "value": value} records in outer-term insertion order. Include a record even when its supplied value is zero. Documents with no contributions have total 0.0 and an empty trace.

Return a dictionary with exactly these keys:

• "totals": a document-to-float mapping in the declared document order;
• "trace": a document-to-list mapping in the same order;
• "ranking": a list of {"document": document, "score": total} records, sorted by descending total, with equal totals kept in declared document order.

Raise ValueError("duplicate document: <id>") for a repeated declared ID, ValueError("unknown document: <id>") for a contribution keyed by an ID not in documents, and ValueError("non-finite contribution: <term>.<id>") for a non-finite value. Keep both input structures unchanged.

Example

For documents ['d1','d2','d3'] and

the totals are d1: 2.0, d2: 0.2, and d3: 1.7; the ranking is d1, d3, d2. The trace for d1 keeps title before body.

Your implementation

Do not import a search library, print, or ask for input. Ordinary dictionaries and lists are sufficient.