# Hpricot + Ultraviolet 0.1.3
#
# See the tests at the bottom for usage examples
#
# Released into the public domain
#
# Please send bug reports and improvements to
# christoffer.sawicki@gmail.com
#
# Noteworthy changes
#  * 0.1.3: Hpricot 0.6 is buggy, use +gem+ to load a older version
#
# Canonical source
#   http://code.vemod.net/svn/misc/hpricot_goodies/hpricot_plus_ultraviolet.rb

if Kernel.methods.include?("gem")
  gem "hpricot", "<0.6" # 0.6 is buggy
end

require "hpricot"
require "uv"

module HpricotPlusUltraviolet
  extend self

  def color_syntax(html, theme = "active4d")
    doc = Hpricot(html)

    strip_unwanted_newlines!(doc)
    
    doc.search("//code[@lang]") do |tag|
      code = CGI.unescapeHTML(tag.inner_html)
      language = tag.attributes.delete("lang")
      
      if Uv.syntaxes.include?(language)
        uv = Uv.parse(code, "xhtml", language, false, theme)
        tag.inner_html = strip_outer_pre_tag(uv)
        tag.attributes["class"] = "ultraviolet"
      end
    end

    return doc.to_html
  end
  
  private
  
  def strip_unwanted_newlines!(doc, tag_names = %w<pre code>)
    # FIXME
    tag_names.each do |tag_name|
      doc.search("//#{tag_name}") do |tag|
        tag.inner_html = tag.inner_html.sub(/\A[\r\n]*/, "").sub(/[\r\n]*\Z/, "")
      end
    end
  end
  
  def strip_outer_pre_tag(html)
    match = html.match(/^<pre.*?>(.*?)<\/pre>$/m)
    
    if match
      match.captures.first
    else
      raise("No <pre> tag to remove in #{html.inspect}")
    end
  end
end

if __FILE__ == $0
  require("test/unit")

  class HpricotPlusUltravioletTest < Test::Unit::TestCase
    def test_stripping_of_newlines
      input = "<pre>\n<code>\nanswer = 42\n</code>\n</pre>"

      expected = "<pre><code>answer = 42</code></pre>"
      actual   = HpricotPlusUltraviolet.color_syntax(input)
    
      assert_equal(expected, actual)
    end
  
    def test_coloring_with_default_theme
      input = "<code lang=\"ruby\">@answer = 42</code>"
    
      expected = "<code class=\"ultraviolet\"><span class=\"Variable\"><span class=\"Variable\">@</span>answer</span> <span class=\"Operator\">=</span> <span class=\"Number\">42</span>\n</code>"
      actual   = HpricotPlusUltraviolet.color_syntax(input)
    
      assert_equal(expected, actual)
    end
  
    def test_coloring_with_brilliance_black_theme
      input = "<code lang=\"ruby\">@answer = 42</code>"
    
      expected = "<code class=\"ultraviolet\"><span class=\"InstanceVariable\"><span class=\"VariablePunctuation\">@</span>answer</span> <span class=\"KeywordOperator\">=</span> <span class=\"Number\">42</span>\n</code>"
      actual   = HpricotPlusUltraviolet.color_syntax(input, "brilliance_black")
    
      assert_equal(expected, actual)
    end
    
    def test_ultraviolet_with_invalid_language
      input = "<code lang=\"cobol\">HELLO WORLD</code>"
      
      expected = "<code>HELLO WORLD</code>"
      actual   = HpricotPlusUltraviolet.color_syntax(input)
    
      assert_equal(expected, actual)
    end
  end
end