1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2022-11-09 12:17:21 -05:00
ruby--ruby/lib/did_you_mean/tree_spell_checker.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

110 lines
2.8 KiB
Ruby
Raw Permalink Normal View History

2020-05-22 17:17:10 -04:00
# frozen_string_literal: true
module DidYouMean
# spell checker for a dictionary that has a tree
# structure, see doc/tree_spell_checker_api.md
class TreeSpellChecker
2020-05-22 17:17:10 -04:00
attr_reader :dictionary, :separator, :augment
def initialize(dictionary:, separator: '/', augment: nil)
@dictionary = dictionary
@separator = separator
@augment = augment
end
def correct(input)
2020-05-22 17:17:10 -04:00
plausibles = plausible_dimensions(input)
return fall_back_to_normal_spell_check(input) if plausibles.empty?
2020-05-22 17:17:10 -04:00
suggestions = find_suggestions(input, plausibles)
return fall_back_to_normal_spell_check(input) if suggestions.empty?
2020-05-22 17:17:10 -04:00
suggestions
end
2020-05-22 17:17:10 -04:00
def dictionary_without_leaves
@dictionary_without_leaves ||= dictionary.map { |word| word.split(separator)[0..-2] }.uniq
end
2020-05-22 17:17:10 -04:00
def tree_depth
@tree_depth ||= dictionary_without_leaves.max { |a, b| a.size <=> b.size }.size
end
2020-05-22 17:17:10 -04:00
def dimensions
@dimensions ||= tree_depth.times.map do |index|
dictionary_without_leaves.map { |element| element[index] }.compact.uniq
end
end
2020-05-22 17:17:10 -04:00
def find_leaves(path)
path_with_separator = "#{path}#{separator}"
dictionary
.select {|str| str.include?(path_with_separator) }
.map {|str| str.gsub(path_with_separator, '') }
end
2020-05-22 17:17:10 -04:00
def plausible_dimensions(input)
input.split(separator)[0..-2]
.map
.with_index { |element, index| correct_element(dimensions[index], element) if dimensions[index] }
.compact
end
def possible_paths(states)
2020-05-22 17:17:10 -04:00
states.map { |state| state.join(separator) }
end
2020-05-22 17:17:10 -04:00
private
2020-05-22 17:17:10 -04:00
def find_suggestions(input, plausibles)
states = plausibles[0].product(*plausibles[1..-1])
paths = possible_paths(states)
leaf = input.split(separator).last
find_ideas(paths, leaf)
end
2020-05-22 17:17:10 -04:00
def fall_back_to_normal_spell_check(input)
return [] unless augment
::DidYouMean::SpellChecker.new(dictionary: dictionary).correct(input)
end
2020-05-22 17:17:10 -04:00
def find_ideas(paths, leaf)
paths.flat_map do |path|
names = find_leaves(path)
ideas = correct_element(names, leaf)
2020-05-22 17:17:10 -04:00
ideas_to_paths(ideas, leaf, names, path)
end.compact
end
2020-05-22 17:17:10 -04:00
def ideas_to_paths(ideas, leaf, names, path)
if ideas.empty?
nil
elsif names.include?(leaf)
["#{path}#{separator}#{leaf}"]
else
ideas.map {|str| "#{path}#{separator}#{str}" }
end
end
2020-05-22 17:17:10 -04:00
def correct_element(names, element)
return names if names.size == 1
2020-05-22 17:17:10 -04:00
str = normalize(element)
2020-05-22 17:17:10 -04:00
return [str] if names.include?(str)
2020-05-22 17:17:10 -04:00
::DidYouMean::SpellChecker.new(dictionary: names).correct(str)
end
2020-05-22 17:17:10 -04:00
def normalize(str)
str.downcase!
2020-05-22 17:17:10 -04:00
str.tr!('@', ' ') if str.include?('@')
str
end
end
end