lib/unibits.rb in unibits-1.3.0 vs lib/unibits.rb in unibits-2.0.0

@@ -2,21 +2,23 @@
 require_relative "unibits/symbolify"
 
 require "io/console"
 require "paint"
 require "unicode/display_width"
-require "unicode/categories"
+require "characteristics"
 
 module Unibits
   SUPPORTED_ENCODINGS = [
     'UTF-8',
     'UTF-16LE',
     'UTF-16BE',
     'UTF-32LE',
     'UTF-32BE',
     'ASCII-8BIT',
     'US-ASCII',
+    /^ISO-8859-/,
+    /^Windows-125/,
   ].freeze
   DEFAULT_TERMINAL_WIDTH = 80
 
   def self.of(string, encoding: nil, convert: nil, stats: true, wide_ambiguous: false, width: nil)
     if !string || string.empty?
@@ -47,47 +49,42 @@
     "\n  #{valid ? '' : Paint["Invalid ", :bold, :red]}#{Paint[string.encoding.name, :bold]} (#{bytes}/#{codepoints}/#{glyphs}/#{width})"
   end
 
   def self.visualize(string, wide_ambiguous: false, width: nil)
     cols = width || determine_terminal_cols
+    encoding_name = string.encoding.name
 
+    type = Characteristics.type_from_encoding_name(encoding_name)
+
     cp_buffer  = ["  "]
     enc_buffer = ["  "]
     hex_buffer = ["  "]
     bin_buffer = ["  "]
     separator  = ["  "]
     current_encoding_error = nil
 
     puts
     string.each_char{ |char|
-      if char.valid_encoding?
-        char_valid = true
-        current_encoding_error = nil
-        if Unicode::Categories.category(char) == "Cn"
-          current_color = "#FF5500"
-        else
-          current_color = random_color
-        end
-      else
-        char_valid = false
-        current_color = :red
-      end
+      char_info = Characteristics.create_for_type(char, type)
+      current_color = determine_char_color(char_info)
 
+      current_encoding_error = nil if char_info.valid?
+
       char.each_byte.with_index{ |byte, index|
         if Paint.unpaint(hex_buffer[-1]).bytesize > cols - 12
           cp_buffer  << "  "
           enc_buffer << "  "
           hex_buffer << "  "
           bin_buffer << "  "
           separator  << "  "
         end
 
         if index == 0
-          if char_valid
+          if char_info.valid?
             codepoint = "U+%04X" % char.ord
           else
-            case string.encoding.name
+            case encoding_name
             when "US-ASCII"
               codepoint = "invalid"
             when "UTF-8"
               # this tries to detect what is wrong with this utf-8 encoded string
               # sorry for this mess
@@ -165,11 +162,11 @@
                 codepoint = "invalid"
               end
             when 'UTF-16LE', 'UTF-16BE'
               if char.bytesize.odd?
                 codepoint = "incompl."
-              elsif char.b[string.encoding.name == 'UTF-16LE' ? 1 : 0].unpack("B*")[0][0, 5] == "11011"
+              elsif char.b[encoding_name == 'UTF-16LE' ? 1 : 0].unpack("B*")[0][0, 5] == "11011"
                 codepoint = "hlf.srg."
               else
                 codepoint = "invalid"
               end
             when 'UTF-32LE', 'UTF-32BE'
@@ -183,18 +180,18 @@
 
           cp_buffer[-1] << Paint[
             codepoint.ljust(10), current_color, :bold
           ]
 
-          if char_valid
-            symbolified_char = symbolify(char)
+          symbolified_char = Symbolify.symbolify(char, char_info)
+
+          if char_info.unicode?
+            padding = 10 - Unicode::DisplayWidth.of(symbolified_char, wide_ambiguous ? 2 : 1)
           else
-            symbolified_char = "�"
+            padding = 10 - symbolified_char.size
           end
 
-          padding = 10 - Unicode::DisplayWidth.of(symbolified_char, wide_ambiguous ? 2 : 1)
-
           enc_buffer[-1] << Paint[
             symbolified_char, current_color
           ]
           enc_buffer[-1] << " " * padding if padding > 0
         else
@@ -206,15 +203,15 @@
           ("%02X" % byte).ljust(10, " "), current_color
         ]
 
         bin_byte_complete = byte.to_s(2).rjust(8, "0")
 
-        if !char_valid
+        if !char_info.valid?
           bin_byte_1 = bin_byte_complete
           bin_byte_2 = ""
         else
-          case string.encoding.name
+          case encoding_name
           when 'US-ASCII'
             bin_byte_1 = bin_byte_complete[0...1]
             bin_byte_2 = bin_byte_complete[1...8]
           when 'ASCII-8BIT'
             bin_byte_1 = ""
@@ -251,10 +248,13 @@
               bin_byte_2 = $2
             end
           when 'UTF-32LE', 'UTF-32BE'
             bin_byte_1 = ""
             bin_byte_2 = bin_byte_complete
+          when /^(ISO-8859-|Windows-125)/
+            bin_byte_1 = ""
+            bin_byte_2 = bin_byte_complete
           end
         end
 
         bin_buffer[-1] << Paint[
           bin_byte_1, current_color
@@ -266,27 +266,36 @@
 
         bin_buffer[-1] << "  "
       }
     }
 
-    if string.encoding.name[0, 3] == "UTF"
+    if type == :unicode
       enc_buffer.zip(cp_buffer, hex_buffer, bin_buffer, separator).flatten.join("\n")
     else
       enc_buffer.zip(hex_buffer, bin_buffer, separator).flatten.join("\n")
     end
   end
 
-  def self.random_color
-    "%.2x%.2x%.2x" %[rand(90) + 60, rand(90) + 60, rand(90) + 60]
-  end
-
-  def self.symbolify(char)
-    return char.inspect unless char.encoding.name[0, 3] == "UTF"
-    Symbolify.symbolify(char).encode('UTF-8')
-  end
-
   def self.determine_terminal_cols
     STDIN.winsize[1] || DEFAULT_TERMINAL_WIDTH
   rescue Errno::ENOTTY
     return DEFAULT_TERMINAL_WIDTH
+  end
+
+  def self.determine_char_color(char_info)
+    if !char_info.valid?
+      "#FF0000"
+    elsif !char_info.assigned?
+      "#FF5500"
+    elsif char_info.control?
+      "#0000FF"
+    elsif char_info.blank?
+      "#33AADD"
+    else
+      random_color
+    end
+  end
+
+  def self.random_color
+    "%.2x%.2x%.2x" % [rand(90) + 60, rand(90) + 60, rand(90) + 60]
   end
 end