Handle force_encoding issue, according to #73

lib/reverse_markdown.rb

@@ -33,15 +33,17 @@
 module ReverseMarkdown
 
   def self.convert(input, options = {})
-    root = case input
-      when String                  then Nokogiri::HTML(input).root
-      when Nokogiri::XML::Document then input.root
-      when Nokogiri::XML::Node     then input
-    end
+    config.with(options) do
+      input = cleaner.force_encoding(input.to_s)
 
-    root or return ''
+      root = case input
+        when String                  then Nokogiri::HTML(input).root
+        when Nokogiri::XML::Document then input.root
+        when Nokogiri::XML::Node     then input
+      end
+
+      root or return ''
 
-    config.with(options) do
       result = ReverseMarkdown::Converters.lookup(root.name).convert(root)
       cleaner.tidy(result)
     end

lib/reverse_markdown/cleaner.rb

@@ -59,6 +59,11 @@ def clean_punctuation_characters(string)
       string.gsub(/(\*\*|~~|__)\s([\.!\?'"])/, "\\1".strip + "\\2")
     end
 
+    def force_encoding(string)
+      ReverseMarkdown.config.force_encoding or return string
+      string.encode('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
+    end
+
     private
 
     def preserve_border_whitespaces(string, options = {}, &block)

lib/reverse_markdown/config.rb

@@ -1,10 +1,11 @@
 module ReverseMarkdown
   class Config
-    attr_accessor :unknown_tags, :github_flavored, :tag_border
+    attr_accessor :unknown_tags, :github_flavored, :tag_border, :force_encoding
 
     def initialize
       @unknown_tags     = :pass_through
       @github_flavored  = false
+      @force_encoding   = false
       @em_delimiter     = '_'.freeze
       @strong_delimiter = '**'.freeze
       @inline_options   = {}
@@ -29,5 +30,9 @@ def github_flavored
     def tag_border
       @inline_options[:tag_border] || @tag_border
     end
+
+    def force_encoding
+      @inline_options[:force_encoding] || @force_encoding
+    end
   end
 end

spec/lib/reverse_markdown_spec.rb

@@ -33,5 +33,15 @@
       end
       expect(ReverseMarkdown.config.github_flavored).to eq true
     end
+
+    describe 'force_encoding option' do
+      it 'raises invalid byte sequence in UTF-8 exception' do
+        expect { ReverseMarkdown.convert("hi \255") }.to raise_error(ArgumentError)
+      end
+
+      it 'handles invalid byte sequence if option is set' do
+        expect(ReverseMarkdown.convert("hi \255", force_encoding: true)).to eq "hi\n\n"
+      end
+    end
   end
 end