Implement detector (#728)

Author: webdev778

ruby/interscript.gemspec

@@ -27,5 +27,6 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
 
   spec.add_dependency "thor"
-  spec.add_dependency "interscript-maps"
+  spec.add_dependency "interscript-maps", "~> #{Interscript::VERSION.split('.')[0,2].join(".")}.0"
+  spec.add_dependency "text"
 end

ruby/lib/interscript.rb

@@ -53,6 +53,16 @@ def transliterate_file(system_code, input_file, output_file, maps={})
       output_file
     end
 
+    # Detects the transliteration that gives the most close approximation
+    # of transliterating source into destination.
+    #
+    # Set multiple: true to get a full report.
+    def detect(source, destination, **kwargs)
+      detector = Detector.new
+      detector.set_from_kwargs(**kwargs)
+      detector.(source, destination)
+    end
+
     def map_gems
       @map_gems ||= Gem.find_latest_files('interscript-maps.yaml').map do |i|
         [i, YAML.load_file(i)]
@@ -109,3 +119,5 @@ def maps(basename: true, load_path: false, select: "*", libraries: false)
 
 require 'interscript/dsl'
 require 'interscript/node'
+
+require 'interscript/detector'

ruby/lib/interscript/detector.rb

@@ -0,0 +1,60 @@
+require "text"
+
+class Interscript::Detector
+  attr_accessor :compiler
+  attr_accessor :distance_computer
+  attr_accessor :map_pattern
+
+  # TODO: use transliterate_each
+  attr_accessor :each
+
+  attr_accessor :load_path
+  attr_accessor :cache
+
+  # Returns a summary of all detected transliterations
+  attr_accessor :multiple
+
+  def initialize
+    @compiler = Interscript::Interpreter
+    @distance_computer = DistanceComputer::Levenshtein
+    @map_pattern = "*"
+
+    @each = false
+
+    @load_path = false
+    @cache = CACHE
+  end
+
+  def set_from_kwargs(**kwargs)
+    kwargs.each do |k,v|
+      self.public_send(:"#{k}=", v)
+    end
+  end
+
+  def call(source, destination)
+    maps = Interscript.maps(select: @map_pattern, load_path: @load_path)
+
+    summary = maps.map do |map|
+      try_dest = Interscript.transliterate(map, source, compiler: @compiler)
+
+      [map, try_dest]
+    end.map do |map, try_dest|
+      dist = @distance_computer.(try_dest, destination)
+      
+      [map, dist]
+    end.sort_by(&:last).to_h
+
+    if @multiple
+      summary.to_h
+    else
+      summary.first.first
+    end
+  end
+
+  CACHE = {}
+
+  # A DistanceComputer needs to respond to #call(source, destination)
+  module DistanceComputer
+    Levenshtein = Text::Levenshtein.method(:distance)
+  end
+end