Add verbose (-v) option

lib/whimsy/utf8-utils.rb

@@ -9,7 +9,7 @@ module UTF8Utils
   #
   # Initially assumes the file is in utf8-softbank encoding
   # If that does not work, then it tries ISO-8859-1
-  def self.repair(src, dst)
+  def self.repair(src, dst, verbose=false)
     opts = {undef: :replace, invalid: :replace}
     ec1 = Encoding::Converter.new('utf8-softbank', "UTF-8", **opts)
     ec2 = Encoding::Converter.new('iso-8859-1', "UTF-8", **opts)
@@ -21,6 +21,11 @@ def self.repair(src, dst)
           if o.include? UTF8_REPLACE # something did not convert
             o = ec2.convert(l) # try another encoding
           end
+          if verbose
+            puts l
+            puts o
+            puts ''
+          end
         end
         w.write o
       end

tools/utf8-fix.rb

@@ -6,8 +6,12 @@
 require 'whimsy/utf8-utils'
 
 if __FILE__ == $0
+  verbose = !ARGV.delete('-v').nil?
   src = ARGV.shift or raise Exception.new "need input file"
   dst = ARGV.shift || src + '.tmp'
-  puts "Input: #{src} output: #{dst}"
-  UTF8Utils::repair(src, dst)
+  puts "Input: #{src} output: #{dst} verbose: #{verbose}"
+  UTF8Utils::repair(src, dst, verbose)
+  if verbose
+    puts "Above are the changed lines. Note that some may appear the same, but the encoding has changed."
+  end
 end