Refactoring

lib/wayback_machine_downloader/to_regex.rb

@@ -1,17 +1,25 @@
+# frozen_string_literal: true
+
 module ToRegex
   module StringMixin
+    INLINE_OPTIONS = /[imxnesu]*/i.freeze
+    REGEXP_DELIMITERS = {
+      '%r{' => '}'.freeze,
+      '/' => '/'.freeze
+    }.freeze
+
+    REGEX_FLAGS = {
+      ignore_case: Regexp::IGNORECASE,
+      multiline: Regexp::MULTILINE,
+      extended: Regexp::EXTENDED
+    }.freeze
+
     class << self
       def literal?(str)
-        REGEXP_DELIMITERS.none? { |s, e| str.start_with?(s) and str =~ /#{e}#{INLINE_OPTIONS}\z/ }
+        REGEXP_DELIMITERS.none? { |start, ending| str.start_with?(start) && str.match?(/#{ending}#{INLINE_OPTIONS}\z/) }
       end
     end
 
-    INLINE_OPTIONS = /[imxnesu]*/
-    REGEXP_DELIMITERS = {
-      '%r{' => '}',
-      '/' => '/',
-    }
-
     # Get a regex back
     #
     # Without :literal or :detect, `"foo".to_regex` will return nil.
@@ -24,54 +32,72 @@ module ToRegex
     # @option options [true,false] :extended /foo/x
     # @option options [true,false] :lang /foo/[nesu]
     def to_regex(options = {})
-      if args = as_regexp(options)
+      args = as_regexp(options)
-        ::Regexp.new(*args)
+      args ? Regexp.new(*args) : nil
-      end
     end
-
     # Return arguments that can be passed to `Regexp.new`
     # @see to_regexp
     def as_regexp(options = {})
-      unless options.is_a?(::Hash)
+      raise ArgumentError, '[to_regexp] Options must be a Hash' unless options.is_a?(Hash)
-        raise ::ArgumentError, "[to_regexp] Options must be a Hash"
+      
-      end
       str = self
+      return if options[:detect] && str.empty?
 
-      return if options[:detect] and str == ''
+      if should_treat_as_literal?(str, options)
-
+        content = Regexp.escape(str)
-      if options[:literal] or (options[:detect] and ToRegexp::String.literal?(str))
+      elsif (delim_set = extract_delimiters(str))
-        content = ::Regexp.escape str
+        content, options = parse_regexp_string(str, delim_set, options)
-      elsif delim_set = REGEXP_DELIMITERS.detect { |k, _| str.start_with?(k) }
+        return unless content
-        delim_start, delim_end = delim_set
-        /\A#{delim_start}(.*)#{delim_end}(#{INLINE_OPTIONS})\z/u =~ str
-        content = $1
-        inline_options = $2
-        return unless content.is_a?(::String)
-        content.gsub! '\\/', '/'
-        if inline_options
-          options[:ignore_case] = true if inline_options.include?('i')
-          options[:multiline] = true if inline_options.include?('m')
-          options[:extended] = true if inline_options.include?('x')
-          # 'n', 'N' = none, 'e', 'E' = EUC, 's', 'S' = SJIS, 'u', 'U' = UTF-8
-          options[:lang] = inline_options.scan(/[nesu]/i).join.downcase
-        end
       else
         return
       end
 
-      ignore_case = options[:ignore_case] ? ::Regexp::IGNORECASE : 0
+      build_regexp_args(content, options)
-      multiline = options[:multiline] ? ::Regexp::MULTILINE : 0
+    end
-      extended = options[:extended] ? ::Regexp::EXTENDED : 0
-      lang = options[:lang] || ''
-      if ::RUBY_VERSION > '1.9' and lang.include?('u')
-        lang = lang.delete 'u'
-      end
 
-      if lang.empty?
+    private
-        [ content, (ignore_case|multiline|extended) ]
+
-      else
+    def should_treat_as_literal?(str, options)
-        [ content, (ignore_case|multiline|extended), lang ]
+      options[:literal] || (options[:detect] && ToRegex::StringMixin.literal?(str))
-      end
+    end
+
+    def extract_delimiters(str)
+      REGEXP_DELIMITERS.find { |start, _| str.start_with?(start) }
+    end
+
+    def parse_regexp_string(str, delim_set, options)
+      start_delim, end_delim = delim_set
+      match = /\A#{start_delim}(.*)#{end_delim}(#{INLINE_OPTIONS})\z/u.match(str)
+      return unless match
+
+      content = match[1].gsub('\\/', '/')
+      parse_inline_options(match[2], options)
+      [content, options]
+    end
+
+    def parse_inline_options(inline_options, options)
+      return unless inline_options
+      options[:ignore_case] = true if inline_options.include?('i')
+      options[:multiline] = true if inline_options.include?('m')
+      options[:extended] = true if inline_options.include?('x')
+      # 'n', 'N' = none, 'e', 'E' = EUC, 's', 'S' = SJIS, 'u', 'U' = UTF-8
+      options[:lang] = inline_options.scan(/[nesu]/i).join.downcase
+    end
+
+    def build_regexp_args(content, options)
+      flags = calculate_flags(options)
+      lang = normalize_lang_option(options[:lang])
+      
+      lang.empty? ? [content, flags] : [content, flags, lang]
+    end
+
+    def calculate_flags(options)
+      REGEX_FLAGS.sum { |key, value| options[key] ? value : 0 }
+    end
+
+    def normalize_lang_option(lang)
+      return '' unless lang
+      RUBY_VERSION >= '1.9' ? lang.delete('u') : lang
     end
   end
 end