Make html2haml usable from withing Ruby, using the Haml::HTML class and/or the Hpricot to_haml extension.

git-svn-id: svn://hamptoncatlin.com/haml/trunk@475 7063305b-7217-0410-af8c-cdc13e5119b9

bin/html2haml

@@ -1,95 +1,6 @@
 #!/usr/bin/env ruby
 
 require File.dirname(__FILE__) + '/../lib/haml'
-require 'haml/engine'
-
-begin
-  require 'rubygems'
-  require 'hpricot'
-rescue LoadError => err
-  dep = err.message.scan(/^no such file to load -- (.*)/)[0]
-  puts "Required dependency #{dep} not found!"
-  exit 1
-end
-
-def tabulate(tabs)
-  '  ' * tabs
-end
-
-TEXT_REGEXP = /^(\s*).*$/
-
-def parse_text(text, tabs)
-  text.strip!
-  if text.empty?
-    String.new
-  else
-    lines = text.split("\n")
-
-    lines.map do |line|
-      line.strip!
-      "#{tabulate(tabs)}#{'\\' if Haml::Engine::SPECIAL_CHARACTERS.include?(line[0])}#{line}\n"
-    end.join
-  end
-end
-
-module Hpricot::Node
-  def to_haml(tabs)
-    parse_text(self.to_s, tabs)
-  end
-end
-
-class Hpricot::Doc
-  def to_haml
-    output = ''
-    children.each { |child| output += child.to_haml(0) }
-    output
-  end
-end
-
-class Hpricot::XMLDecl
-  def to_haml(tabs)
-    "#{tabulate(tabs)}!!! XML\n"
-  end
-end
-
-class Hpricot::DocType
-  def to_haml(tabs)
-    "#{tabulate(tabs)}!!!\n"
-  end
-end
-
-class Hpricot::Comment
-  def to_haml(tabs)
-    "#{tabulate(tabs)}/\n#{parse_text(self.content, tabs + 1)}"
-  end
-end
-
-class Hpricot::Elem
-  def to_haml(tabs)
-    output = "#{tabulate(tabs)}"
-    output += "%#{name}" unless name == 'div' && (attributes.include?('id') || attributes.include?('class'))
-      
-    if attributes
-      output += "##{attributes['id']}" if attributes['id']
-      attributes['class'].split(' ').each { |c| output += ".#{c}" } if attributes['class']
-      attributes.delete("id")
-      attributes.delete("class")
-      output += attributes.inspect if attributes.length > 0
-    end
-  
-    output += "/" if children.length == 0
-    output += "\n"
-
-    self.children.each do |child|
-      output += child.to_haml(tabs + 1)
-    end
-
-    output
-  end
-end
-
-# Must be required after Hpricot mods,
-# so they're in scope
 require 'haml/exec'
 
 opts = Haml::Exec::HTML2Haml.new(ARGV)

lib/haml/exec.rb

@@ -220,6 +220,18 @@ END
     # A class encapsulating executable functionality
     # specific to the html2haml executable.
     class HTML2Haml < Generic # :nodoc:
+      def initialize(args)
+        super
+
+        begin
+          require 'haml/html'
+        rescue LoadError => err
+          dep = err.message.scan(/^no such file to load -- (.*)/)[0]
+          puts "Required dependency #{dep} not found!"
+          exit 1
+        end
+      end
+
       def set_opts(opts)
         opts.banner = <<END
 Usage: html2haml [options] (html file) (output file)
@@ -238,7 +250,7 @@ END
         input = @options[:input]
         output = @options[:output]
 
-        output.write(Hpricot(input).to_haml)
+        output.write(::Haml::HTML.new(input).render)
       end
     end
   end

lib/haml/html.rb

@@ -0,0 +1,114 @@
+require File.dirname(__FILE__) + '/../haml'
+
+require 'haml/engine'
+require 'rubygems'
+require 'hpricot'
+
+module Haml
+  # This class contains the functionality used in the +html2haml+ utility,
+  # namely converting HTML documents to Haml templates.
+  # It depends on Hpricot for HTML parsing (http://code.whytheluckystiff.net/hpricot/).
+  class HTML
+    # Creates a new instance of Haml::HTML that will compile the given template,
+    # which can either be a string containing HTML or an Hpricot node,
+    # to a Haml string when +render+ is called.
+    def initialize(template)
+      if template.is_a? Hpricot::Node
+        @template = template
+      else
+        @template = Hpricot(template)
+      end
+    end
+
+    # Processes the document and returns the result as a string
+    # containing the Haml template.
+    def render
+      @template.to_haml(0)
+    end
+    alias_method :to_haml, :render
+
+    module ::Hpricot::Node
+      # Returns the Haml representation of the given node,
+      # at the given tabulation.
+      def to_haml(tabs = 0)
+        parse_text(self.to_s, tabs)
+      end
+
+      private
+
+      def tabulate(tabs)
+        '  ' * tabs
+      end
+
+      def parse_text(text, tabs)
+        text.strip!
+        if text.empty?
+          String.new
+        else
+          lines = text.split("\n")
+          
+          lines.map do |line|
+            line.strip!
+            "#{tabulate(tabs)}#{'\\' if Haml::Engine::SPECIAL_CHARACTERS.include?(line[0])}#{line}\n"
+          end.join
+        end
+      end
+    end
+
+    # :stopdoc:
+    
+    TEXT_REGEXP = /^(\s*).*$/
+
+    class ::Hpricot::Doc
+      def to_haml(tabs = 0)
+        output = ''
+        children.each { |child| output += child.to_haml(0) }
+        output
+      end
+    end
+
+    class ::Hpricot::XMLDecl
+      def to_haml(tabs = 0)
+        "#{tabulate(tabs)}!!! XML\n"
+      end
+    end
+
+    # TODO: Understand STRICT, etc. doctypes
+    class ::Hpricot::DocType
+      def to_haml(tabs = 0)
+        "#{tabulate(tabs)}!!!\n"
+      end
+    end
+
+    class ::Hpricot::Comment
+      def to_haml(tabs = 0)
+        "#{tabulate(tabs)}/\n#{parse_text(self.content, tabs + 1)}"
+      end
+    end
+
+    class ::Hpricot::Elem
+      def to_haml(tabs = 0)
+        output = "#{tabulate(tabs)}"
+        output += "%#{name}" unless name == 'div' && (attributes.include?('id') || attributes.include?('class'))
+        
+        if attributes
+          output += "##{attributes['id']}" if attributes['id']
+          attributes['class'].split(' ').each { |c| output += ".#{c}" } if attributes['class']
+          attributes.delete("id")
+          attributes.delete("class")
+          output += attributes.inspect if attributes.length > 0
+        end
+        
+        output += "/" if children.length == 0
+        output += "\n"
+        
+        self.children.each do |child|
+          output += child.to_haml(tabs + 1)
+        end
+
+        output
+      end
+    end
+    # :startdoc:
+  end
+end