Sha256: f222bdf6427dc27c24d30f7b8a4c8d4303dac540eaa242a15578b6c5f0e72e79
Contents?: true
Size: 1.29 KB
Versions: 8
Compression:
Stored size: 1.29 KB
Contents
# frozen-string-literal: true
require "did_you_mean/levenshtein"
require "did_you_mean/jaro_winkler"
module DidYouMean
class SpellChecker
def initialize(dictionary: )
@dictionary = dictionary
end
def correct(input)
input = normalize(input)
threshold = input.length > 3 ? 0.834 : 0.77
words = @dictionary.select {|word| JaroWinkler.distance(normalize(word), input) >= threshold }
words.reject! {|word| input == word.to_s }
words.sort_by! {|word| JaroWinkler.distance(word.to_s, input) }
words.reverse!
# Correct mistypes
threshold = (input.length * 0.25).ceil
corrections = words.select {|c| Levenshtein.distance(normalize(c), input) <= threshold }
# Correct misspells
if corrections.empty?
corrections = words.select do |word|
word = normalize(word)
length = input.length < word.length ? input.length : word.length
Levenshtein.distance(word, input) < length
end.first(1)
end
corrections
end
private
def normalize(str_or_symbol) #:nodoc:
str = if str_or_symbol.is_a?(String)
str_or_symbol.dup
else
str_or_symbol.to_s
end
str.downcase!
str.tr!("@", "")
str
end
end
end
Version data entries
8 entries across 8 versions & 2 rubygems