lib/regextest/front/letter.rb in regextest-0.1.5 vs lib/regextest/front/letter.rb in regextest-0.1.6

@@ -36,21 +36,26 @@
         @data_type = :LEX_CHAR
         @obj = CharClass.new([ TRange.new(val)])
       when :LEX_SIMPLE_ESCAPE
         @data_type = :LEX_CHAR
         @obj = CharClass.new([ TRange.new(val[1..1])])
-      when :LEX_CODE_LITERAL, :LEX_ESCAPED_LETTER, :LEX_UNICODE, :LEX_CONTROL_LETTER, :LEX_META_LETTER, :LEX_OCTET
+      when :LEX_CODE_LITERAL, :LEX_ESCAPED_LETTER, :LEX_UNICODE, :LEX_OCTET
         @data_type = :LEX_CHAR
         @obj = CharClass.new([ TRange.new(eval('"'+ val + '"'))])   # convert using ruby's eval
+      when :LEX_CONTROL_LETTER, :LEX_META_LETTER
+        @data_type = :LEX_CHAR
+        @obj = generate_control_letter(val, type)
       when :LEX_BRACKET
         @obj = Regextest::Front::Bracket.new(val)
       when :LEX_SIMPLIFIED_CLASS
         @obj = generate_simplified_class(val)
       when :LEX_POSIX_CHAR_CLASS
         @obj = generate_char_class(val)
       when :LEX_UNICODE_CLASS
-        @obj = generate_unicode_char(val)
+        @obj = generate_unicode_char(val, type)
+      when :LEX_UNICODE_CLASS_BRACKET
+        @obj = generate_unicode_char(val, type)
       when :LEX_ANY_LETTER
         @obj = generate_any_char(val)
       when :LEX_SPECIAL_LETTER
         @obj = generate_special_char(val)
       when :LEX_AND_AND
@@ -59,18 +64,44 @@
       else
         raise "Error: internal error, type:#{type} not implemented"
       end
     end
     
+    # generate control letter \c-x, \m-x
+    def generate_control_letter(val, type)
+      suffix = val[-1..-1]
+      codepoint = suffix.unpack("U*")[0]
+      case type
+      when :LEX_CONTROL_LETTER
+        if    ('0'..'?').include?(suffix)
+          result = codepoint - 0x20
+        elsif ('@'..'_').include?(suffix)
+          result = codepoint - 0x40
+        elsif ('`'..'~').include?(suffix)
+          result = codepoint - 0x60
+        else
+          raise "Internal error: invalid control letter (#{val})"
+        end
+      when :LEX_META_LETTER
+        result = codepoint + 0x80
+        pp [result].pack("U*")
+      else
+        raise "Internal error: invalid type #{type}"
+      end
+      @obj = CharClass.new([ TRange.new([result].pack("U*"))])
+    end
+    
     # generate whole set of letters (depends on option)
     def generate_any_char(val)
       if @options[:reg_options].is_unicode?
         obj = CharClass.new(TstConstUnicodeCharSet)
       else
-        obj = CharClass.new( [ TRange.new("\x20", "\x7e") ] )
+        obj = CharClass.new(TstConstUnicodeCharSet)
+        # obj = CharClass.new( [ TRange.new("\x20", "\x7e") ] )
       end
       
+      # add new-line if multi-line option specified
       if( @options[:reg_options].is_multiline? )
           obj.add_ranges( [ TRange.new("\n") ] )
       end
       obj
     end
@@ -175,16 +206,16 @@
       end
       obj
     end
     
     # generate Unicode class (ie. \p{...} | \P{...})
-    def generate_unicode_char(val)
+    def generate_unicode_char(val, type)
       if(md = val.match(/(p|P)\{(\^?)(\w+)\}/))
         class_name = md[3].downcase
         reverse = (md[2] && md[2]=="^")?true:false
         
-        obj = CharClass.new(class_name)
+        obj = CharClass.new(class_name, type)
       else
         raise "Internal error, inconsistent Unicode class #{val}"
       end
       
       # ￥P{^...} is equivalent to \p{...}
@@ -295,12 +326,14 @@
     end
     
     # transform to json format
     def json
       @@id += 1
+      charset = @options[:reg_options].charset
       "{" +
         "\"type\": \"#{@data_type}\", \"id\": \"L#{@@id}\", \"value\": #{@obj.json}, " +
-         "\"offset\": #{@offset}, \"length\": #{@length}" +
+         "\"offset\": #{@offset}, \"length\": #{@length}, " +
+         "\"charset\": \"#{charset}\"" +
       "}"
     end
   end
 end