Work on html2haml, including a reasonable UI and ability to escape Haml-sensitive plain text.

git-svn-id: svn://hamptoncatlin.com/haml/trunk@336 7063305b-7217-0410-af8c-cdc13e5119b9

bin/html2haml

@@ -1,22 +1,27 @@
 #!/usr/bin/env ruby
 
+require File.dirname(__FILE__) + '/../lib/haml'
+require 'haml/engine'
 require 'rubygems'
 require 'hpricot'
-require 'open-uri'
-
-doc = open(ARGV[0]) { |f| Hpricot(f) }
 
 def tabulate(tabs)
   '  ' * tabs
 end
 
+TEXT_REGEXP = /^(\s*).*$/
+
 def parse_text(text, tabs)
   text.strip!
   if text.empty?
     String.new
   else
-    text = text.gsub(/[\t ]+/, " ").gsub(/(\r|\n|\r\n)\s*/, "\n#{tabulate(tabs)}")
-    "#{tabulate(tabs)}#{text}\n"
+    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
 
@@ -76,5 +81,9 @@ class Hpricot::Elem
   end
 end
 
-puts doc.to_haml
+# Must be required after Hpricot mods,
+# so they're in scope
+require 'haml/exec'
 
+opts = Haml::Exec::HTML2Haml.new(ARGV)
+opts.parse!

lib/haml/exec.rb

@@ -95,7 +95,7 @@ module Haml
 
       def set_opts(opts)
         opts.banner = <<END
-Usage: #{@name.downcase} [options] (template file) (output file)
+Usage: #{@name.downcase} [options] (#{@name.downcase} file) (output file)
 
 Description:
   Uses the #{@name} engine to parse the specified template
@@ -154,5 +154,30 @@ END
         output.close() if output.is_a? File
       end
     end
+
+    # A class encapsulating executable functionality
+    # specific to the html2haml executable.
+    class HTML2Haml < Generic
+      def set_opts(opts)
+        opts.banner = <<END
+Usage: html2haml [options] (html file) (output file)
+
+Description: Transforms an HTML file into corresponding Haml code.
+
+Options:
+END
+
+        super
+      end
+
+      def process_result
+        super
+
+        input = @options[:input]
+        output = @options[:output]
+
+        output.write(Hpricot(input).to_haml)
+      end
+    end
   end
 end