Idea: use raw text windows for fuzzy match auto-resolution #35
Description
Idea (to explore)
When fuzzy matching a character name, before prompting the user, scan all source files for occurrences of the candidate token. Extract the following word — only if capitalized (potential last name, filters out verbs/punctuation noise).
If all capitalized-next-word occurrences agree on the same token (e.g. always "Baggins") → only one entity, auto-resolve without asking.
If multiple last names found (e.g. "Baggins" + "Took") → ambiguous, fall back to user prompt.
Limits
- Only works for characters with consistent first+last name in the text
- Single-name characters (Elias, Jakes...) won't benefit
- Heuristic, not guaranteed
Where to plug in
Somewhere before the user confirmation step in the fuzzy match flow.