# Search mentities based on a `query` string
	#
	# Lookup the index for anything matching `query` and return `limit` results.
	#
	# The algorithm used is the following:
	# 1- lookup by name prefix
	# 2- lookup by full_name prefix
	# 3- loopup by levenshtein distance
	#
	# At each step if the `limit` is reached, the algorithm stops and returns the results.
	fun find(query: String, limit: nullable Int, filter: nullable ModelFilter): Array[MEntity] do
		# Find, lookup by name prefix
		var matches = index.find_by_name_prefix(query, filter).uniq.
			sort(lname_sorter, name_sorter, kind_sorter)
		if limit != null and matches.length >= limit then
			return matches.limit(limit).rerank.sort(vis_sorter, score_sorter).mentities
		end
		matches = matches.rerank.sort(vis_sorter, score_sorter)

		# If limit not reached, lookup by full_name prefix
		var malus = matches.length
		var full_matches = new IndexMatches
		for match in index.find_by_full_name_prefix(query, filter).
			sort(kind_sorter, lfname_sorter, fname_sorter) do
			match.score += malus
			full_matches.add match
		end
		matches = matches.uniq
		if limit != null and matches.length + full_matches.length >= limit then
			matches.add_all full_matches
			matches = matches.uniq.limit(limit).rerank.sort(vis_sorter, score_sorter)
			return matches.mentities
		end

		# If limit not reached, lookup by similarity
		malus = matches.length
		var sim_matches = new IndexMatches
		for match in index.find_by_similarity(query, filter).sort(score_sorter, kind_sorter, lname_sorter, name_sorter) do
			match.score += malus
			sim_matches.add match
		end
		matches.add_all sim_matches
		matches = matches.uniq
		if limit != null then matches = matches.limit(limit)
		return matches.rerank.sort(vis_sorter, score_sorter).mentities
	end