Make Parser a class and remove from Engine

You can now instantiate your own Haml parser separately from the Haml
engine if for some reason you should want to. Eventually this will be
useful for tools that convert Haml to Erb, for example. However for the
moment the primary motivation is simply to guarantee a stricter separation
of concerns between the parser, compiler and engine.

lib/haml/engine.rb

@@ -16,7 +16,6 @@ module Haml
   #     output = haml_engine.render
   #     puts output
   class Engine
-    include Parser
     include Compiler
 
     # The options hash.
@@ -70,6 +69,23 @@ module Haml
       end
     end
 
+
+    DEFAULT_OPTIONS = {
+      :suppress_eval        => false,
+      :attr_wrapper         => "'",
+      # Don't forget to update the docs in doc-src/REFERENCE.md
+      # if you update these
+      :autoclose            => %w[meta img link br hr input area param col base],
+      :preserve             => %w[textarea pre code],
+      :filename             => '(haml)',
+      :line                 => 1,
+      :ugly                 => false,
+      :format               => :xhtml,
+      :escape_html          => false,
+      :escape_attrs         => true,
+      :hyphenate_data_attrs => true,
+    }
+
     # Precompiles the Haml template.
     #
     # @param template [String] The Haml template
@@ -77,31 +93,15 @@ module Haml
     #   see {file:REFERENCE.md#options the Haml options documentation}
     # @raise [Haml::Error] if there's a Haml syntax error in the template
     def initialize(template, options = {})
-      @options = {
-        :suppress_eval        => false,
-        :attr_wrapper         => "'",
-        # Don't forget to update the docs in doc-src/REFERENCE.md
-        # if you update these
-        :autoclose            => %w[meta img link br hr input area param col base],
-        :preserve             => %w[textarea pre code],
-        :filename             => '(haml)',
-        :line                 => 1,
-        :ugly                 => false,
-        :format               => :xhtml,
-        :escape_html          => false,
-        :escape_attrs         => true,
-        :hyphenate_data_attrs => true,
-      }
-
       @index = nil # explicitily initialize to avoid warnings
+      @options = DEFAULT_OPTIONS.dup.merge! options.reject {|k, v| v.nil?}
 
-      template = check_haml_encoding(template) do |msg, line|
+      @template = check_haml_encoding(template) do |msg, line|
         raise Haml::Error.new(msg, line)
       end
 
-      set_up_encoding(options, template)
+      set_up_encoding
 
-      @options.merge! options.reject {|k, v| v.nil?}
       @index = 0
 
       @options[:format] = :xhtml if @options[:mime_type] == 'text/xml'
@@ -110,19 +110,13 @@ module Haml
         raise Haml::Error, "Invalid output format #{@options[:format].inspect}"
       end
 
-      # :eod is a special end-of-document marker
+      @parser = Parser.new(@template, @options)
-      @template = (template.rstrip).split(/\r\n|\r|\n/) + [:eod, :eod]
-      @template_index = 0
-      @to_close_stack = []
-      @output_tabs = 0
-      @template_tabs = 0
-      @flat = false
-      @newlines = 0
-      @precompiled = ''
-      @to_merge = []
-      @tab_change  = 0
 
-      compile(parse)
+      @output_tabs = 0
+      @to_merge    = []
+      @precompiled = ''
+
+      compile(@parser.parse)
     rescue Haml::Error => e
       if @index || e.line
         e.backtrace.unshift "#{@options[:filename]}:#{(e.line ? e.line + 1 : @index) + @options[:line] - 1}"
@@ -307,15 +301,19 @@ module Haml
     private
 
     if RUBY_VERSION < "1.9"
-      def set_up_encoding(options, template)
+      def set_up_encoding
-        options
       end
     else
-      def set_up_encoding(options, template)
+      def set_up_encoding
         @options.tap do |ops|
-          ops[:encoding] = Encoding.default_internal || template.encoding
+          ops[:encoding] ||= Encoding.default_internal || @template.encoding
-          ops[:encoding] = "utf-8" if ops[:encoding].name == "US-ASCII"
+          if ops[:encoding].is_a?(Encoding)
-          ops[:encoding] = ops[:encoding].name if ops[:encoding].is_a?(Encoding)
+            if ops[:encoding].name == "US-ASCII"
+              ops[:encoding] = "utf-8"
+            else
+              ops[:encoding] = ops[:encoding].name
+            end
+          end
         end
       end
     end

lib/haml/html.rb

@@ -78,7 +78,7 @@ module Hpricot
 
       text.split("\n").map do |line|
         line.strip!
-        "#{tabulate(tabs)}#{'\\' if Haml::Engine::SPECIAL_CHARACTERS.include?(line[0])}#{line}\n"
+        "#{tabulate(tabs)}#{'\\' if Haml::Parser::SPECIAL_CHARACTERS.include?(line[0])}#{line}\n"
       end.join
     end
   end

lib/haml/parser.rb

@@ -1,7 +1,7 @@
 require 'strscan'
 
 module Haml
-  module Parser
+  class Parser
     include Haml::Util
 
     # Designates an XHTML/XML element.
@@ -75,6 +75,54 @@ module Haml
     # The Regex that matches a literal string or symbol value
     LITERAL_VALUE_REGEX = /:(\w*)|(["'])((?![\\#]|\2).|\\.)*\2/
 
+    def initialize(template, options)
+      # :eod is a special end-of-document marker
+      @template       = (template.rstrip).split(/\r\n|\r|\n/) + [:eod, :eod]
+      @options        = options
+      @flat           = false
+      @template_index = 0
+      @template_tabs  = 0
+    end
+
+    def parse
+      @root = @parent = ParseNode.new(:root)
+      @haml_comment = false
+      @indentation = nil
+      @line = next_line
+
+      raise SyntaxError.new("Indenting at the beginning of the document is illegal.", @line.index) if @line.tabs != 0
+
+      while next_line
+        process_indent(@line) unless @line.text.empty?
+
+        if flat?
+          text = @line.full.dup
+          text = "" unless text.gsub!(/^#{@flat_spaces}/, '')
+          @filter_buffer << "#{text}\n"
+          @line = @next_line
+          next
+        end
+
+        @tab_up = nil
+        process_line(@line.text, @line.index) unless @line.text.empty? || @haml_comment
+        if @parent.type != :haml_comment && (block_opened? || @tab_up)
+          @template_tabs += 1
+          @parent = @parent.children.last
+        end
+
+        if !@haml_comment && !flat? && @next_line.tabs - @line.tabs > 1
+          raise SyntaxError.new("The line was indented #{@next_line.tabs - @line.tabs} levels deeper than the previous line.", @next_line.index)
+        end
+
+        @line = @next_line
+      end
+
+      # Close all the open tags
+      close until @parent.type == :root
+      @root
+    end
+
+
     private
 
     # @private
@@ -124,44 +172,6 @@ END
       end
     end
 
-    def parse
-      @root = @parent = ParseNode.new(:root)
-      @haml_comment = false
-      @indentation = nil
-      @line = next_line
-
-      raise SyntaxError.new("Indenting at the beginning of the document is illegal.", @line.index) if @line.tabs != 0
-
-      while next_line
-        process_indent(@line) unless @line.text.empty?
-
-        if flat?
-          text = @line.full.dup
-          text = "" unless text.gsub!(/^#{@flat_spaces}/, '')
-          @filter_buffer << "#{text}\n"
-          @line = @next_line
-          next
-        end
-
-        @tab_up = nil
-        process_line(@line.text, @line.index) unless @line.text.empty? || @haml_comment
-        if @parent.type != :haml_comment && (block_opened? || @tab_up)
-          @template_tabs += 1
-          @parent = @parent.children.last
-        end
-
-        if !@haml_comment && !flat? && @next_line.tabs - @line.tabs > 1
-          raise SyntaxError.new("The line was indented #{@next_line.tabs - @line.tabs} levels deeper than the previous line.", @next_line.index)
-        end
-
-        @line = @next_line
-      end
-
-      # Close all the open tags
-      close until @parent.type == :root
-      @root
-    end
-
     # Processes and deals with lowering indentation.
     def process_indent(line)
       return unless line.tabs <= @template_tabs && @template_tabs > 0
@@ -675,26 +685,6 @@ END
         !((text[-3..-2] =~ /\W\?/) || text[-3..-2] == "?\\")
     end
 
-    def contains_interpolation?(str)
-      str.include?('#{')
-    end
-
-    def unescape_interpolation(str, escape_html = nil)
-      res = ''
-      rest = Haml::Util.handle_interpolation str.dump do |scan|
-        escapes = (scan[2].size - 1) / 2
-        res << scan.matched[0...-3 - escapes]
-        if escapes % 2 == 1
-          res << '#{'
-        else
-          content = eval('"' + balance(scan, ?{, ?}, 1)[0][0...-1] + '"')
-          content = "Haml::Helpers.html_escape((#{content}))" if escape_html
-          res << '#{' + content + "}"# Use eval to get rid of string escapes
-        end
-      end
-      res + rest
-    end
-
     def balance(*args)
       res = Haml::Util.balance(*args)
       return res if res

lib/haml/util.rb

@@ -319,6 +319,26 @@ METHOD
       "#{indentation.length} #{noun}#{'s' unless singular}#{was}"
     end
 
+    def contains_interpolation?(str)
+      str.include?('#{')
+    end
+
+    def unescape_interpolation(str, escape_html = nil)
+      res = ''
+      rest = Haml::Util.handle_interpolation str.dump do |scan|
+        escapes = (scan[2].size - 1) / 2
+        res << scan.matched[0...-3 - escapes]
+        if escapes % 2 == 1
+          res << '#{'
+        else
+          content = eval('"' + balance(scan, ?{, ?}, 1)[0][0...-1] + '"')
+          content = "Haml::Helpers.html_escape((#{content}))" if escape_html
+          res << '#{' + content + "}"# Use eval to get rid of string escapes
+        end
+      end
+      res + rest
+    end
+
     private
 
     # Parses a magic comment at the beginning of a Haml file.