[HTML search] correction for scoring of search terms that only appear in document indexed with id zero.

[jayaddison]

Feature or Bugfix

• Bugfix

Purpose

• Restores expected scoring for exact-match query terms that only appear in document zero in the client-side JavaScript search index.
• Includes a performance optimization: skips partial-matching entirely when exact-matches exist.

Detail

• Words (terms) from both document content and document titles are stored in the JavaScript search index on the client -- this results in a mapping from individual terms to a list of the document-ids of the documents where they appear.
• For terms/title-terms that appear in multiple documents, the document-ids are stored as a (sorted) list/Array. In cases where the term only appears in a single document, only a single integer value is stored. (credit to @wlach for explaining)
• The document-ids are zero-indexed (0, 1, 2, ...) so there is a zeroth document, and we need to be careful when evaluating that in boolean expressions because zero is falsy in JavaScript.
• 🐛 A term that only appears in the document with ID zero will incorrectly be scored as a partial-match.

Relates

• Resolves HTML Search: Partial matches can overwrite main matches in certain rare cases #11957.
• Resolves [HTML search] optimization: don't loop over all document terms and title terms during search. #12045.
• Alternative approach to HTML Search: Fix partial matches overwriting full matches #11958.

cc @wlach

[jayaddison]

Replaced by / became-identical-to #11957.