Give a more descriptive error for templates with invalid byte sequences in 1.9.

doc-src/HAML_CHANGELOG.md

@@ -17,6 +17,12 @@ in the generated class and id.
 The Haml executable now has a `--double-quote-attributes` option (short form: `-q`)
 that causes attributes to use a double-quote mark rather than single-quote.
 
+### Ruby 1.9 Support
+
+Haml now produces more descriptive errors when given a template
+with invalid byte sequences for that template's encoding,
+including the line number and the offending character.
+
 ## [2.2.4](http://github.com/nex3/haml/commit/2.2.4)
 
 * Allow `end` to be used for silent script when it's followed by code.

doc-src/SASS_CHANGELOG.md

@@ -16,6 +16,12 @@ In addition, when the `sass` executable encounters an error,
 it now prints the filename where the error occurs,
 as well as a backtrace of Sass imports.
 
+### Ruby 1.9 Support
+
+Sass now produces more descriptive errors when given a template
+with invalid byte sequences for that template's encoding,
+including the line number and the offending character.
+
 ### `css2sass` Error Handling
 
 Several bug fixes and minor improvements have been made, including:

lib/haml/engine.rb

@@ -96,6 +96,8 @@ module Haml
         @options[:encoding] = @options[:encoding].name
       end
 
+      check_encoding(template) {|msg, line| raise Haml::Error.new(msg, line)}
+
       # :eod is a special end-of-document marker
       @template = (template.rstrip).split(/\r\n|\r|\n/) + [:eod, :eod]
       @template_index = 0
@@ -111,7 +113,9 @@ module Haml
 
       precompile
     rescue Haml::Error => e
-      e.backtrace.unshift "#{@options[:filename]}:#{(e.line ? e.line + 1 : @index) + @options[:line] - 1}" if @index
+      if @index || e.line
+        e.backtrace.unshift "#{@options[:filename]}:#{(e.line ? e.line + 1 : @index) + @options[:line] - 1}"
+      end
       raise
     end
 

lib/haml/util.rb

@@ -129,6 +129,22 @@ module Haml
       Haml::Util::RUBY_VERSION[0] == 1 && Haml::Util::RUBY_VERSION[1] < 9
     end
 
+    def check_encoding(str)
+      return if ruby1_8?
+      return if str.valid_encoding?
+      encoding = str.encoding
+      newlines = Regexp.new("\r\n|\r|\n".encode(encoding).force_encoding("binary"))
+      str.force_encoding("binary").split(newlines).each_with_index do |line, i|
+        begin
+          line.encode(encoding)
+        rescue Encoding::UndefinedConversionError => e
+          yield <<MSG.rstrip, i + 1
+Invalid #{encoding.name} character #{e.error_char.dump}
+MSG
+        end
+      end
+    end
+
     # Checks to see if a class has a given method.
     # For example:
     #

lib/sass/engine.rb

@@ -156,6 +156,8 @@ module Sass
     # @return [Sass::Tree::Node] The root of the parse tree.
     # @raise [Sass::SyntaxError] if there's an error in the document
     def to_tree
+      check_encoding(@template) {|msg, line| raise Sass::SyntaxError.new(msg, :line => line)}
+
       root = Tree::RootNode.new(@template)
       append_children(root, tree(tabulate(@template)).first, true)
       root.options = @options

test/haml/engine_test.rb

@@ -1061,6 +1061,24 @@ HAML
         o.render
       end
     end
+
+    def test_encoding_error
+      render("foo\nbar\nb\xFEaz".force_encoding("utf-8"))
+      assert(false, "Expected exception")
+    rescue Haml::Error => e
+      assert_equal(3, e.line)
+      assert_equal('Invalid UTF-8 character "\xFE"', e.message)
+    end
+
+    def test_ascii_incompatible_encoding_error
+      template = "foo\nbar\nb_z".encode("utf-16le")
+      template[9] = "\xFE".force_encoding("utf-16le")
+      render(template)
+      assert(false, "Expected exception")
+    rescue Haml::Error => e
+      assert_equal(3, e.line)
+      assert_equal('Invalid UTF-16LE character "\xFE"', e.message)
+    end
   end
 
   private

test/sass/engine_test.rb

@@ -901,6 +901,28 @@ a
 SASS
   end
 
+  # Encodings
+
+  unless Haml::Util.ruby1_8?
+    def test_encoding_error
+      render("foo\nbar\nb\xFEaz".force_encoding("utf-8"))
+      assert(false, "Expected exception")
+    rescue Sass::SyntaxError => e
+      assert_equal(3, e.sass_line)
+      assert_equal('Invalid UTF-8 character "\xFE"', e.message)
+    end
+
+    def test_ascii_incompatible_encoding_error
+      template = "foo\nbar\nb_z".encode("utf-16le")
+      template[9] = "\xFE".force_encoding("utf-16le")
+      render(template)
+      assert(false, "Expected exception")
+    rescue Sass::SyntaxError => e
+      assert_equal(3, e.sass_line)
+      assert_equal('Invalid UTF-16LE character "\xFE"', e.message)
+    end
+  end
+
   private
   
   def render(sass, options = {})