Fuzzy process extractone

Process.ExtractOne

The single best candidate, or nothing.

public static ExtractResult? ExtractOne(string query, IEnumerable<string> choices, Func<string, string, double> scorer = null, double scoreCutoff = 0)

Parameters — query is what to match. choices are the candidates. scorer defaults to Fuzz.WRatio. scoreCutoff is the score a candidate must clear.

Returns — ExtractResult? — the best candidate, or null when none clears scoreCutoff.

Example — a match, and a query that clears nothing.

using Lodestar.Fuzzy;

string[] choices = ["apple pie", "apple tart", "banana bread", "cherry pie"];

ExtractResult? found = Process.ExtractOne("appel pie", choices);
string choice = found?.Choice;  // => apple pie

ExtractResult? none = Process.ExtractOne("zzz", choices, scoreCutoff: 90);
bool nothingMatched = none is null;  // => True

Remarks — the nullable return is the point. With scoreCutoff at its default of 0 every query matches something, and the best of four unrelated candidates is still returned — which is how a fuzzy match quietly becomes a wrong answer. Setting a cutoff makes "none of these" an outcome the type can express, and null is that outcome.

Choosing the cutoff is a judgement about the data rather than a default worth shipping: 90 suits near-identical strings, and far lower suits free text.

Applies to — net10.0, netstandard2.0.

See also — Process.Extract, ExtractResult.