unibits.rb in unibits-2.2.0

- old
+ new

@@ -8,10 +8,11 @@
 
 module Unibits
   SUPPORTED_ENCODINGS = Encoding.name_list.grep(
     Regexp.union(
       /^UTF-8$/,
+      /^UTF8-/,
       /^UTF-...E$/,
       /^ASCII-8BIT$/,
       /^US-ASCII$/,
       /^ISO-8859-/,
       /^Windows-125/,
@@ -28,10 +29,11 @@
     invalid: "#FF0000",
     unassigned: "#FF5500",
     control: "#0000FF",
     blank: "#33AADD",
     format: "#FF00FF",
+    mark: "#228822",
   }
 
   DEFAULT_TERMINAL_WIDTH = 80
 
   def self.of(string, encoding: nil, convert: nil, stats: true, wide_ambiguous: false, width: nil)
@@ -98,11 +100,11 @@
             codepoint = "U+%04X" % char.ord
           else
             case encoding_name
             when "US-ASCII"
               codepoint = "invalid"
-            when "UTF-8"
+            when "UTF-8", /^UTF8/
               # this tries to detect what is wrong with this utf-8 encoded string
               # sorry for this mess
               case char.unpack("B*")[0]
               when /^110.{5}$/
                 current_encoding_error = [:nec, 1, 1]
@@ -242,10 +244,12 @@
       COLORS[:control]
     elsif char_info.blank?
       COLORS[:blank]
     elsif char_info.format?
       COLORS[:format]
+    elsif char_info.unicode? && char_info.category[0] == "M"
+      COLORS[:mark]
     else
       random_color
     end
   end
 
@@ -265,10 +269,10 @@
         bin_byte_1 = bin_byte_complete[0...1]
         bin_byte_2 = bin_byte_complete[1...8]
       when 'ASCII-8BIT'
         bin_byte_1 = ""
         bin_byte_2 = bin_byte_complete
-      when 'UTF-8'
+      when 'UTF-8', /^UTF8/
         if byteindex == 0
           if bin_byte_complete =~ /^(0|1{2,4}0)([01]+)$/
             bin_byte_1 = $1
             bin_byte_2 = $2
           else