fast-fuzzy

Fast fuzzy search utility

methodology

fast-fuzzy is an on-line fuzzy searching utility that is pretty fast for moderate sized lists of candidates. The ranking algorithm is a modification of levenshtein distance proposed by Peter H. Sellers (paper). We also use the damerau-levenshtein distance as opposed to normal levenshtein in order to get better results.

Inputs are normalized by taking the lowercase of a string, removing non-word characters, reducing all whitespace to single spaces, and trimming off any leading or trailing whitespace, and calling string.normalize.

exports

• fuzzy - the core fuzzy search ranking function. returns a number between 0 and 1, with 1 being perfect match and 0 being perfect mismatch.
• search - search function for one-off searches. returns a filtered and sorted copy of the original array.
• Searcher - a class for searching the same set of candidates multiple times. caches normalization and key selection.

options

Searcher and search both take an options object for configuring behavior.

• keySelector - a function which selects the string to search (if candidates are objects). Default is (_) => _
• threshold - a number from 0 to 1 representing the minimum score. Default is .6
• ignoreCase - a boolean specifying whether or not to ignore case. Default is true
• ignoreSymbols - a boolean specifying whether or not to ignore `~!@#$%^&*()-=_+{}[]|\\;':",./<>?. Default is true
• normalizeWhitespace - a boolean specifying whether or not to normalize and trim whitespace. Default is true
• returnScores - a boolean specifying whether or not to return the scores. Default is false
  • objects are returned in the form {item, key, score}
• useDamerau - a boolean specifying whether or not to use the damerau-levenshtein distance. Default is true

fuzzy accepts a subset of these options (ignoreCase, ignoreSymbols, normalizeWhitespace, useDamerau) with the same defaults.