viktorasl/fuzzysearch

Lightweight Fuzzy evaluation protocol with CollectionType extension

FuzzySearch

Lightweight Fuzzy evaluation protocol with CollectionType extension

Requirements

iOS 8.0+ Swift 3.0

Usage

Implementing FuzzySearchable protocol

struct PlayerModel {
  let name: String
  let position: String
  let goals: Int
}

Implementation of FuzzySearchable protocol defines against what search patterns will be evaluated.

extension PlayerModel: FuzzySearchable {
  var fuzzyStringToMatch: String {
    return name
  }
}

Evaluating single FuzzySearchable instance

let maradona = PlayerModel(name: "Diego Maradona", position: "F", goals: 16)
maradona.fuzzyMatch("diema") // FuzzySearchResult(weight: 15, parts: [(0,3), (6,2)])

FuzzySearchResult

Result of evaluation carries two properties:

• weight - weight of the match
• parts - NSRange's of pattern matching against fuzzyStringToMatch

Evaluating collection of FuzzySearchables

When evaluating collection of FuzzySearchables result is an array of tuples (item: Generator.Element, result: FuzzySearchResult) which is filtered and sorted depending on weight.

let players = [
  PlayerModel(name: "Diego Maradona", position: "CF", goals: 16),
  PlayerModel(name: "David Beckham", position: "CAM", goals: 8),
  PlayerModel(name: "Lionel Messi", position: "RW", goals: 15)
]
players.fuzzyMatch("di")
//[
// (
//  FuzzySearchTests.PlayerModel(name: "Diego Maradona", position: "CF", goals: 16),
//  FuzzySearch.FuzzySearchResult(weight: 4, parts: [(0,2)])
// ), (
//  FuzzySearchTests.PlayerModel(name: "David Beckham", position: "CAM", goals: 8),
//  FuzzySearch.FuzzySearchResult(weight: 2, parts: [(0,1), (3,1)])
// )
//]

CachedFuzzySearchable<T: FuzzySearchable>

Wraps over a FuzzySearchable instance, caching some underlying metadata generated while fuzzy-matching w/ FuzzySearchable.fuzzyMatch.

Use this cached wrapper over FuzzySearchable instances that are expected to be fuzzy-matched multiple times without mutation to fuzzyStringToMatch:

let players = [
  PlayerModel(name: "Diego Maradona", position: "CF", goals: 16),
  PlayerModel(name: "David Beckham", position: "CAM", goals: 8),
  PlayerModel(name: "Lionel Messi", position: "RW", goals: 15),
  // Many more players ...
]

let fuzzyCachedPlayers =
  players.map { player in 
    CachedFuzzySearchable(wrapping: player) 
  }

fuzzyCachedPlayers.fuzzyMatch("di")

// Subsequente calls to 'fuzzyMatch' over the array above cause less overhead when re-matching
fuzzyCachedPlayers.fuzzyMatch("di") // Runs in about half time

players.fuzzyMatch("di") // Will fuzzy-match against original, non-cached PlayerModel values.

CachedFuzzySearchable has storage needs of a little over 2x that of the fuzzyStringToSearch property of the wrapped FuzzySearchable value. Discarding a CachedFuzzySearchable value also discards the extra memory that was allocated.

Returning a different value from the wrapped FuzzySearchable's fuzzyStringToSearch property resets the cache automatically during the next fuzzyMatch call and the overhead is reset to that of a fresh CachedFuzzySearchable instance, so implementers shouldn't worry about mantaining a stable fuzzyStringToSearch return.

License

FuzzySearch is released under the MIT license.