regru-premailer 1.7.4

data/lib/premailer/adapter/nokogiri.rb

@@ -0,0 +1,221 @@
+require 'nokogiri'
+
+class Premailer
+  module Adapter
+    # Nokogiri adapter
+    module Nokogiri
+
+      # Merge CSS into the HTML document.
+      #
+      # @return [String] an HTML.
+      def to_inline_css
+        doc = @processed_doc
+        @unmergable_rules = CssParser::Parser.new
+
+        # Give all styles already in style attributes a specificity of 1000
+        # per http://www.w3.org/TR/CSS21/cascade.html#specificity
+        doc.search("*[@style]").each do |el|
+          el['style'] = '[SPEC=1000[' + el.attributes['style'] + ']]'
+        end
+
+        # Iterate through the rules and merge them into the HTML
+        @css_parser.each_selector(:all) do |selector, declaration, specificity|
+          # Save un-mergable rules separately
+          selector.gsub!(/:link([\s]*)+/i) {|m| $1 }
+
+          # Convert element names to lower case
+          selector.gsub!(/([\s]|^)([\w]+)/) {|m| $1.to_s + $2.to_s.downcase }
+
+          if selector =~ Premailer::RE_UNMERGABLE_SELECTORS
+            @unmergable_rules.add_rule_set!(CssParser::RuleSet.new(selector, declaration)) unless @options[:preserve_styles]
+          else
+            begin
+              # Change single ID CSS selectors into xpath so that we can match more
+              # than one element.  Added to work around dodgy generated code.
+              selector.gsub!(/\A\#([\w_\-]+)\Z/, '*[@id=\1]')
+
+              doc.search(selector).each do |el|
+                if el.elem? and (el.name != 'head' and el.parent.name != 'head')
+                  # Add a style attribute or append to the existing one
+                  block = "[SPEC=#{specificity}[#{declaration}]]"
+                  el['style'] = (el.attributes['style'].to_s ||= '') + ' ' + block
+                end
+              end
+            rescue  ::Nokogiri::SyntaxError, RuntimeError, ArgumentError
+              $stderr.puts "CSS syntax error with selector: #{selector}" if @options[:verbose]
+              next
+            end
+          end
+        end
+
+        # Remove script tags
+        if @options[:remove_scripts]
+          doc.search("script").remove
+        end
+
+        # Read STYLE attributes and perform folding
+        doc.search("*[@style]").each do |el|
+          style = el.attributes['style'].to_s
+
+          declarations = []
+          style.scan(/\[SPEC\=([\d]+)\[(.[^\]\]]*)\]\]/).each do |declaration|
+            rs = CssParser::RuleSet.new(nil, declaration[1].to_s, declaration[0].to_i)
+            declarations << rs
+          end
+
+          # Perform style folding
+          merged = CssParser.merge(declarations)
+          merged.expand_shorthand!
+
+          # Duplicate CSS attributes as HTML attributes
+          if Premailer::RELATED_ATTRIBUTES.has_key?(el.name)
+            Premailer::RELATED_ATTRIBUTES[el.name].each do |css_att, html_att|
+              el[html_att] = merged[css_att].gsub(/url\('(.*)'\)/,'\1').gsub(/;$/, '').strip if el[html_att].nil? and not merged[css_att].empty?
+            end
+          end
+
+          # write the inline STYLE attribute
+          el['style'] = Premailer.escape_string(merged.declarations_to_s).split(';').map(&:strip).sort.join('; ')
+        end
+
+        doc = write_unmergable_css_rules(doc, @unmergable_rules)
+
+        if @options[:remove_classes] or @options[:remove_comments]
+          doc.traverse do |el|
+            if el.comment? and @options[:remove_comments]
+              el.remove
+            elsif el.element?
+              el.remove_attribute('class') if @options[:remove_classes]
+            end
+          end
+        end
+
+        if @options[:remove_ids]
+          # find all anchor's targets and hash them
+          targets = []
+          doc.search("a[@href^='#']").each do |el|
+            target = el.get_attribute('href')[1..-1]
+            targets << target
+            el.set_attribute('href', "#" + Digest::MD5.hexdigest(target))
+          end
+          # hash ids that are links target, delete others
+          doc.search("*[@id]").each do |el|
+            id = el.get_attribute('id')
+            if targets.include?(id)
+              el.set_attribute('id', Digest::MD5.hexdigest(id))
+            else
+              el.remove_attribute('id')
+            end
+          end
+        end
+
+        @processed_doc = doc
+        if is_xhtml?
+          # we don't want to encode carriage returns
+          @processed_doc.to_xhtml(:encoding => nil).gsub(/&\#xD;/i, "\r")
+        else
+          @processed_doc.to_html
+        end
+      end
+
+      # Create a <tt>style</tt> element with un-mergable rules (e.g. <tt>:hover</tt>)
+      # and write it into the <tt>body</tt>.
+      #
+      # <tt>doc</tt> is an Nokogiri document and <tt>unmergable_css_rules</tt> is a Css::RuleSet.
+      #
+      # @return [::Nokogiri::XML] a document.
+      def write_unmergable_css_rules(doc, unmergable_rules) # :nodoc:
+        styles = ''
+        unmergable_rules.each_selector(:all, :force_important => true) do |selector, declarations, specificity|
+          styles += "#{selector} { #{declarations} }\n"
+        end
+        
+        unless styles.empty?
+          style_tag = "<style type=\"text/css\">\n#{styles}></style>"
+          if body = doc.search('body')
+            doc.at_css('body').children.before(::Nokogiri::XML.fragment(style_tag))            
+          else
+            doc.inner_html = style_tag += doc.inner_html
+          end
+        end
+        doc
+      end
+
+
+      # Converts the HTML document to a format suitable for plain-text e-mail.
+      #
+      # If present, uses the <body> element as its base; otherwise uses the whole document.
+      #
+      # @return [String] a plain text.
+      def to_plain_text
+        html_src = ''
+        begin
+          html_src = @doc.at("body").inner_html
+        rescue; end
+
+        html_src = @doc.to_html unless html_src and not html_src.empty?
+        convert_to_text(html_src, @options[:line_length], @html_encoding)
+      end
+
+      # Gets the original HTML as a string.
+      # @return [String] HTML.
+      def to_s
+        if is_xhtml?
+          @doc.to_xhtml(:encoding => nil)
+        else
+          @doc.to_html(:encoding => nil)
+        end
+      end
+
+      # Load the HTML file and convert it into an Nokogiri document.
+      #
+      # @return [::Nokogiri::XML] a document.
+      def load_html(input) # :nodoc:
+        thing = nil
+
+        # TODO: duplicate options
+        if @options[:with_html_string] or @options[:inline] or input.respond_to?(:read)
+          thing = input
+				elsif @is_local_file
+          @base_dir = File.dirname(input)
+          thing = File.open(input, 'r')
+				else
+          thing = open(input)
+        end
+
+        if thing.respond_to?(:read)
+          thing = thing.read
+        end
+
+        return nil unless thing
+        doc = nil
+
+        # Handle HTML entities
+        if @options[:replace_html_entities] == true and thing.is_a?(String) 
+          if RUBY_VERSION =~ /1.9/
+            html_entity_ruby_version = "1.9"
+          elsif RUBY_VERSION =~ /1.8/
+            html_entity_ruby_version = "1.8"
+          end
+          if html_entity_ruby_version
+            HTML_ENTITIES[html_entity_ruby_version].map do |entity, replacement|
+              thing.gsub! entity, replacement
+            end
+          end
+        end
+        # Default encoding is ASCII-8BIT (binary) per http://groups.google.com/group/nokogiri-talk/msg/0b81ef0dc180dc74
+        # However, we really don't want to hardcode this. ASCII-8BIG should be the default, but not the only option.
+        if thing.is_a?(String) and RUBY_VERSION =~ /1.9/
+          thing = thing.force_encoding(@options[:input_encoding]).encode!
+          doc = ::Nokogiri::HTML(thing, nil, @options[:input_encoding]) {|c| c.recover }
+        else
+          default_encoding = RUBY_PLATFORM == 'java' ? nil : 'BINARY'
+          doc = ::Nokogiri::HTML(thing, nil, @options[:input_encoding] || default_encoding) {|c| c.recover }
+        end
+
+        return doc
+      end
+
+    end
+  end
+end

data/lib/premailer/executor.rb

@@ -0,0 +1,100 @@
+require 'optparse'
+require 'premailer'
+
+# defaults
+options = {
+  :base_url => nil,
+  :link_query_string => nil,
+  :remove_classes => false,
+  :verbose => false,
+  :line_length => 65
+}
+
+mode = :html
+
+opts = OptionParser.new do |opts|
+  opts.banner = "Improve the rendering of HTML emails by making CSS inline among other things. Takes a path to a local file, a URL or a pipe as input.\n\n"
+  opts.define_head "Usage: premailer <optional uri|optional path> [options]"
+  opts.separator ""
+  opts.separator "Examples:"
+  opts.separator "  premailer http://example.com/ > out.html"
+  opts.separator "  premailer http://example.com/ --mode txt > out.txt"
+  opts.separator "  cat input.html | premailer -q src=email > out.html"
+  opts.separator "  premailer ./public/index.html"
+  opts.separator ""
+  opts.separator "Options:"
+
+  opts.on("--mode MODE", [:html, :txt], "Output: html or txt") do |v|
+    mode = v
+  end
+
+  opts.on("-b", "--base-url STRING", String, "Base URL, useful for local files") do |v|
+    options[:base_url] = v
+  end
+
+  opts.on("-q", "--query-string STRING", String, "Query string to append to links") do |v|
+    options[:link_query_string] = v
+  end
+
+  opts.on("--css FILE,FILE", Array, "Additional CSS stylesheets") do |v|
+    options[:css] = v
+  end
+
+  opts.on("-r", "--remove-classes", "Remove HTML classes") do |v|
+    options[:remove_classes] = v
+  end
+
+  opts.on("-j", "--remove-scripts", "Remove <script> elements") do |v|
+    options[:remove_classes] = v
+  end
+
+  opts.on("-l", "--line-length N", Integer, "Line length for plaintext (default: #{options[:line_length].to_s})") do |v|
+    options[:line_length] = v
+  end
+
+  opts.on("-d", "--io-exceptions", "Abort on I/O errors") do |v|
+    options[:io_exceptions] = v
+  end
+
+  opts.on("-v", "--verbose", "Print additional information at runtime") do |v|
+    options[:verbose] = v
+  end
+
+  opts.on_tail("-?", "--help", "Show this message") do
+    puts opts
+    exit
+  end
+
+  opts.on_tail("-V", "--version", "Show version") do
+    puts "Premailer #{Premailer::VERSION} (c) 2008-2010 Alex Dunae"
+    exit
+  end
+end
+opts.parse!
+
+$stderr.puts "Processing in #{mode} mode with options #{options.inspect}" if options[:verbose]
+
+premailer = nil
+input = nil
+
+if $stdin.tty? or STDIN.fcntl(Fcntl::F_GETFL, 0) == 0
+  input = ARGV.shift
+else
+  input = $stdin
+  options[:with_html_string] = true
+end
+
+if input
+  premailer = Premailer.new(input, options)
+else
+  puts opts
+  exit 1
+end
+
+if mode == :txt
+  print premailer.to_plain_text
+else
+  print premailer.to_inline_css
+end
+
+exit

data/lib/premailer/html_to_plain_text.rb

@@ -0,0 +1,105 @@
+# coding: utf-8
+require 'htmlentities'
+
+# Support functions for Premailer
+module HtmlToPlainText
+
+  # Returns the text in UTF-8 format with all HTML tags removed
+  #
+  # TODO: add support for DL, OL
+  def convert_to_text(html, line_length = 65, from_charset = 'UTF-8')
+    txt = html
+
+    # decode HTML entities
+    he = HTMLEntities.new
+    txt = he.decode(txt)
+
+    # replace image by their alt attribute
+    txt.gsub!(/<img.+?alt=\"([^\"]*)\"[^>]*\/>/i, '\1')
+
+    # replace image by their alt attribute
+    txt.gsub!(/<img.+?alt=\"([^\"]*)\"[^>]*\/>/i, '\1')
+    txt.gsub!(/<img.+?alt='([^\']*)\'[^>]*\/>/i, '\1')
+
+    # links
+    txt.gsub!(/<a.+?href=\"(mailto:)?([^\"]*)\"[^>]*>((.|\s)+?)<\/a>/i) do |s|
+      $3.strip + ' ( ' + $2.strip + ' )'
+    end
+
+    txt.gsub!(/<a.+?href='(mailto:)?([^\']*)\'[^>]*>((.|\s)+?)<\/a>/i) do |s|
+      $3.strip + ' ( ' + $2.strip + ' )'
+    end
+
+
+    # handle headings (H1-H6)
+    txt.gsub!(/(<\/h[1-6]>)/i, "\n\\1") # move closing tags to new lines
+    txt.gsub!(/[\s]*<h([1-6]+)[^>]*>[\s]*(.*)[\s]*<\/h[1-6]+>/i) do |s|
+      hlevel = $1.to_i
+
+      htext = $2
+      htext.gsub!(/<br[\s]*\/?>/i, "\n") # handle <br>s
+      htext.gsub!(/<\/?[^>]*>/i, '') # strip tags
+
+      # determine maximum line length
+      hlength = 0
+      htext.each_line { |l| llength = l.strip.length; hlength = llength if llength > hlength }
+      hlength = line_length if hlength > line_length
+
+      case hlevel
+        when 1   # H1, asterisks above and below
+          htext = ('*' * hlength) + "\n" + htext + "\n" + ('*' * hlength)
+        when 2   # H1, dashes above and below
+          htext = ('-' * hlength) + "\n" + htext + "\n" + ('-' * hlength)
+        else     # H3-H6, dashes below
+          htext = htext + "\n" + ('-' * hlength)
+      end
+
+      "\n\n" + htext + "\n\n"
+    end
+
+    # wrap spans
+    txt.gsub!(/(<\/span>)[\s]+(<span)/mi, '\1 \2')
+
+    # lists -- TODO: should handle ordered lists
+    txt.gsub!(/[\s]*(<li[^>]*>)[\s]*/i, '* ')
+    # list not followed by a newline
+    txt.gsub!(/<\/li>[\s]*(?![\n])/i, "\n")
+
+    # paragraphs and line breaks
+    txt.gsub!(/<\/p>/i, "\n\n")
+    txt.gsub!(/<br[\/ ]*>/i, "\n")
+
+    # strip remaining tags
+    txt.gsub!(/<\/?[^>]*>/, '')
+
+    txt = word_wrap(txt, line_length)
+
+    # remove linefeeds (\r\n and \r -> \n)
+    txt.gsub!(/\r\n?/, "\n")
+
+    # strip extra spaces
+    txt.gsub!(/\302\240+/, " ") # non-breaking spaces -> spaces
+    txt.gsub!(/\n[ \t]+/, "\n") # space at start of lines
+    txt.gsub!(/[ \t]+\n/, "\n") # space at end of lines
+
+    # no more than two consecutive newlines
+    txt.gsub!(/[\n]{3,}/, "\n\n")
+
+    # no more than two consecutive spaces
+    txt.gsub!(/ {2,}/, " ")
+
+    # the word messes up the parens
+    txt.gsub!(/\([ \n](http[^)]+)[\n ]\)/) do |s|
+      "( " + $1 + " )"
+    end
+
+    txt.strip
+  end
+
+  # Taken from Rails' word_wrap helper (http://api.rubyonrails.org/classes/ActionView/Helpers/TextHelper.html#method-i-word_wrap)
+  def word_wrap(txt, line_length)
+    txt.split("\n").collect do |line|
+      line.length > line_length ? line.gsub(/(.{1,#{line_length}})(\s+|$)/, "\\1\n").strip : line
+    end * "\n"
+  end
+end

data/lib/premailer/premailer.rb

@@ -0,0 +1,549 @@
+# Premailer processes HTML and CSS to improve e-mail deliverability.
+#
+# Premailer's main function is to render all CSS as inline <tt>style</tt>
+# attributes. It also converts relative links to absolute links and checks
+# the 'safety' of CSS properties against a CSS support chart.
+#
+# ## Example of use
+#
+# ```ruby
+# premailer = Premailer.new('http://example.com/myfile.html', :warn_level => Premailer::Warnings::SAFE)
+#
+# # Write the HTML output
+# fout = File.open("output.html", "w")
+# fout.puts premailer.to_inline_css
+# fout.close
+#
+# # Write the plain-text output
+# fout = File.open("ouput.txt", "w")
+# fout.puts premailer.to_plain_text
+# fout.close
+#
+# # List any CSS warnings
+# puts premailer.warnings.length.to_s + ' warnings found'
+# premailer.warnings.each do |w|
+#   puts "#{w[:message]} (#{w[:level]}) may not render properly in #{w[:clients]}"
+# end
+#
+# premailer = Premailer.new(html_file, :warn_level => Premailer::Warnings::SAFE)
+# puts premailer.to_inline_css
+# ```
+#
+class Premailer
+  include HtmlToPlainText
+  include CssParser
+
+  # Premailer version.
+  VERSION = '1.7.4'
+
+  CLIENT_SUPPORT_FILE = File.dirname(__FILE__) + '/../../misc/client_support.yaml'
+
+  # Unmergable selectors regexp.
+  RE_UNMERGABLE_SELECTORS = /(\:(visited|active|hover|focus|after|before|selection|target|first\-(line|letter))|^\@)/i
+  # Reset selectors regexp.
+  RE_RESET_SELECTORS = /^(\:\#outlook|body.*|\.ReadMsgBody|\.ExternalClass|img|\#backgroundTable)$/
+
+  # list of HTMLEntities to fix
+  # source: http://stackoverflow.com/questions/2812781/how-to-convert-webpage-apostrophe-8217-to-ascii-39-in-ruby-1-
+  HTML_ENTITIES = {
+    "1.8" => {
+      "\342\200\231" => "'",
+      "\342\200\246" => "...",
+      "\342\200\176" => "'",
+      "\342\200\177" => "'",
+      "\342\200\230" => "'",
+      "\342\200\231" => "'",
+      "\342\200\232" => ',',
+      "\342\200\233" => "'",
+      "\342\200\234" => '"',
+      "\342\200\235" => '"',
+      "\342\200\041" => '-',
+      "\342\200\174" => '-',
+      "\342\200\220" => '-',
+      "\342\200\223" => '-',
+      "\342\200\224" => '--',
+      "\342\200\225" => '--',
+      "\342\200\042" => '--'
+    },
+    "1.9" => {
+      "&#8217;" => "'",
+      "&#8230;" => "...",
+      "&#8216;" => "'",
+      "&#8218;" => ',',
+      "&#8219;" => "'",
+      "&#8220;" => '"',
+      "&#8221;" => '"',
+      "&#8208;" => '-',
+      "&#8211;" => '-',
+      "&#8212;" => '--',
+      "&#8213;" => '--'
+    }
+  }
+
+  # list of CSS attributes that can be rendered as HTML attributes
+  #
+  # @todo too much repetition
+  # @todo background=""
+  RELATED_ATTRIBUTES = {
+    'h1' => {'text-align' => 'align'},
+    'h2' => {'text-align' => 'align'},
+    'h3' => {'text-align' => 'align'},
+    'h4' => {'text-align' => 'align'},
+    'h5' => {'text-align' => 'align'},
+    'h6' => {'text-align' => 'align'},
+    'p' => {'text-align' => 'align'},
+    'div' => {'text-align' => 'align'},
+    'blockquote' => {'text-align' => 'align'},
+    'body' => {'background-color' => 'bgcolor'},
+    'table' => {
+      'background-color' => 'bgcolor',
+      'background-image' => 'background',
+      '-premailer-width' => 'width',
+      '-premailer-height' => 'height',
+      '-premailer-cellpadding' => 'cellpadding',
+      '-premailer-cellspacing' => 'cellspacing',
+    },
+    'tr' => {
+      'text-align' => 'align',
+      'background-color' => 'bgcolor',
+      '-premailer-height' => 'height'
+    },
+    'th' => {
+      'text-align' => 'align',
+      'background-color' => 'bgcolor',
+      'vertical-align' => 'valign',
+      '-premailer-width' => 'width',
+      '-premailer-height' => 'height'
+    },
+    'td' => {
+      'text-align' => 'align',
+      'background-color' => 'bgcolor',
+      'vertical-align' => 'valign',
+      '-premailer-width' => 'width',
+      '-premailer-height' => 'height'
+    },
+    'img' => {'float' => 'align'}
+  }
+
+  # URI of the HTML file used
+  attr_reader   :html_file
+
+  # base URL used to resolve links
+  attr_reader   :base_url
+
+  # base directory used to resolve links for local files
+  # @return [String] base directory
+  attr_reader   :base_dir
+
+  # unmergeable CSS rules to be preserved in the head (CssParser)
+  attr_reader   :unmergable_rules
+
+  # processed HTML document (Hpricot/Nokogiri)
+  attr_reader   :processed_doc
+
+  # source HTML document (Hpricot/Nokogiri)
+  attr_reader   :doc
+
+  # Warning levels
+  module Warnings
+    # No warnings
+    NONE = 0
+    # Safe
+    SAFE = 1
+    # Poor
+    POOR = 2
+    # Risky
+    RISKY = 3
+  end
+  include Warnings
+
+  # Waning level names
+  WARN_LABEL = %w(NONE SAFE POOR RISKY)
+
+  # Create a new Premailer object.
+  #
+  # @param html is the HTML data to process. It can be either an IO object, the URL of a
+  #   remote file, a local path or a raw HTML string.  If passing an HTML string you
+  #   must set the with_html_string option to true.
+  #
+  # @param [Hash] options the options to handle html with.
+  # @option options [FixNum] :line_length Line length used by to_plain_text. Default is 65.
+  # @option options [FixNum] :warn_level What level of CSS compatibility warnings to show (see {Premailer::Warnings}).
+  # @option options [String] :link_query_string A string to append to every <tt>a href=""</tt> link. Do not include the initial <tt>?</tt>.
+  # @option options [String] :base_url Used to calculate absolute URLs for local files.
+  # @option options [Array(String)] :css Manually specify CSS stylesheets.
+  # @option options [Boolean] :css_to_attributes Copy related CSS attributes into HTML attributes (e.g. background-color to bgcolor)
+  # @option options [String] :css_string Pass CSS as a string
+  # @option options [Boolean] :remove_ids Remove ID attributes whenever possible and convert IDs used as anchors to hashed to avoid collisions in webmail programs.  Default is false.
+  # @option options [Boolean] :remove_classes Remove class attributes. Default is false.
+  # @option options [Boolean] :remove_comments Remove html comments. Default is false.
+  # @option options [Boolean] :remove_scripts Remove <tt>script</tt> elements. Default is true.
+  # @option options [Boolean] :preserve_styles Whether to preserve any <tt>link rel=stylesheet</tt> and <tt>style</tt> elements.  Default is false.
+  # @option options [Boolean] :preserve_reset Whether to preserve styles associated with the MailChimp reset code.
+  # @option options [Boolean] :with_html_string Whether the html param should be treated as a raw string.
+  # @option options [Boolean] :verbose Whether to print errors and warnings to <tt>$stderr</tt>.  Default is false.
+  # @option options [Boolean] :include_link_tags Whether to include css from <tt>link rel=stylesheet</tt> tags.  Default is true.
+  # @option options [Boolean] :include_style_tags Whether to include css from <tt>style</tt> tags.  Default is true.
+  # @option options [String] :input_encoding Manually specify the source documents encoding. This is a good idea.
+  # @option options [Boolean] :replace_html_entities Convert HTML entities to actual characters. Default is false.
+  # @option options [Symbol] :adapter Which HTML parser to use, either <tt>:nokogiri</tt> or <tt>:hpricot</tt>.  Default is <tt>:hpricot</tt>.
+  def initialize(html, options = {})
+    @options = {:warn_level => Warnings::SAFE,
+                :line_length => 65,
+                :link_query_string => nil,
+                :base_url => nil,
+                :remove_classes => false,
+                :remove_ids => false,
+                :remove_comments => false,
+                :remove_scripts => true,
+                :css => [],
+                :css_to_attributes => true,
+                :with_html_string => false,
+                :css_string => nil,
+                :preserve_styles => false,
+                :preserve_reset => true,
+                :verbose => false,
+                :debug => false,
+                :io_exceptions => false,
+                :include_link_tags => true,
+                :include_style_tags => true,
+                :input_encoding => 'ASCII-8BIT',
+                :replace_html_entities => false,
+                :adapter => Adapter.use,
+                }.merge(options)
+
+    @html_file = html
+    @is_local_file = @options[:with_html_string] || Premailer.local_data?(html)
+
+    @css_files = [@options[:css]].flatten
+
+    @css_warnings = []
+
+    @base_url = nil
+    @base_dir = nil
+    @unmergable_rules = nil
+
+    if @options[:base_url]
+      @base_url = URI.parse(@options.delete(:base_url))
+    elsif not @is_local_file
+      @base_url = URI.parse(@html_file)
+    end
+
+    @css_parser = CssParser::Parser.new({
+      :absolute_paths => true,
+      :import => true,
+      :io_exceptions => @options[:io_exceptions]
+    })
+
+    @adapter_class = Adapter.find @options[:adapter]
+
+    self.class.send(:include, @adapter_class)
+
+    @doc = load_html(@html_file)
+
+    @processed_doc = @doc
+    @processed_doc = convert_inline_links(@processed_doc, @base_url) if @base_url
+    if options[:link_query_string]
+      @processed_doc = append_query_string(@processed_doc, options[:link_query_string])
+    end
+    load_css_from_options!
+    load_css_from_html!
+  end
+
+  # CSS warnings.
+  # @return [Array(Hash)] Array of warnings.
+  def warnings
+    return [] if @options[:warn_level] == Warnings::NONE
+    @css_warnings = check_client_support if @css_warnings.empty?
+    @css_warnings
+  end
+
+protected
+  def load_css_from_local_file!(path)
+    css_block = ''
+    begin
+      File.open(path, "r") do |file|
+        while line = file.gets
+          css_block << line
+        end
+      end
+
+      load_css_from_string(css_block)
+    rescue; end
+  end
+
+  def load_css_from_string(css_string)
+    @css_parser.add_block!(css_string, {:base_uri => @base_url, :base_dir => @base_dir, :only_media_types => [:screen, :handheld]})
+  end
+
+  # @private
+  def load_css_from_options! # :nodoc:
+    load_css_from_string(@options[:css_string]) if @options[:css_string]
+
+    @css_files.each do |css_file|
+      if Premailer.local_data?(css_file)
+        load_css_from_local_file!(css_file)
+      else
+        @css_parser.load_uri!(css_file)
+      end
+    end
+  end
+
+    # Load CSS included in <tt>style</tt> and <tt>link</tt> tags from an HTML document.
+  def load_css_from_html! # :nodoc:
+    if tags = @doc.search("link[@rel='stylesheet'], style")
+      tags.each do |tag|
+        if tag.to_s.strip =~ /^\<link/i && tag.attributes['href'] && media_type_ok?(tag.attributes['media']) && @options[:include_link_tags]
+          # A user might want to <link /> to a local css file that is also mirrored on the site
+          # but the local one is different (e.g. newer) than the live file, premailer will now choose the local file
+          
+          if tag.attributes['href'].to_s.include? @base_url.to_s and @html_file.kind_of?(String)
+            link_uri = File.join(File.dirname(@html_file), tag.attributes['href'].to_s.sub!(@base_url.to_s, ''))
+          end
+          
+          # if the file does not exist locally, try to grab the remote reference
+          if link_uri.nil? or not File.exists?(link_uri)
+            link_uri = Premailer.resolve_link(tag.attributes['href'].to_s, @html_file)
+          end
+          
+          if Premailer.local_data?(link_uri)
+            $stderr.puts "Loading css from local file: " + link_uri if @options[:verbose]
+            load_css_from_local_file!(link_uri)
+          else
+            $stderr.puts "Loading css from uri: " + link_uri if @options[:verbose]
+            @css_parser.load_uri!(link_uri, {:only_media_types => [:screen, :handheld]})
+          end
+
+        elsif tag.to_s.strip =~ /^\<style/i && @options[:include_style_tags]
+          @css_parser.add_block!(tag.inner_html, :base_uri => @base_url, :base_dir => @base_dir, :only_media_types => [:screen, :handheld])
+        end
+      end
+      tags.remove unless @options[:preserve_styles]
+    end
+  end
+
+
+
+# here be deprecated methods
+public
+  # @private
+  # @deprecated
+  def local_uri?(uri) # :nodoc:
+    warn "[DEPRECATION] `local_uri?` is deprecated.  Please use `Premailer.local_data?` instead."
+    Premailer.local_data?(uri)
+  end
+
+# here be instance methods
+
+  # @private
+  def media_type_ok?(media_types)
+    return true if media_types.nil? or media_types.empty?
+    media_types.split(/[\s]+|,/).any? { |media_type| media_type.strip =~ /screen|handheld|all/i }
+  rescue
+    true
+  end
+
+  def append_query_string(doc, qs)
+    return doc if qs.nil?
+
+    qs.to_s.gsub!(/^[\?]*/, '').strip!
+    return doc if qs.empty?
+
+    begin
+      current_host = @base_url.host
+    rescue
+      current_host = nil
+    end
+
+    $stderr.puts "Attempting to append_query_string: #{qs}" if @options[:verbose]
+
+    doc.search('a').each do|el|
+      href = el.attributes['href'].to_s.strip
+      next if href.nil? or href.empty?
+      
+      next if href[0,1] =~ /[\#\{\[\<\%]/ # don't bother with anchors or special-looking links
+
+      begin
+        href = URI.parse(href)
+
+        if current_host and href.host != nil and href.host != current_host
+          $stderr.puts "Skipping append_query_string for: #{href.to_s} because host is no good" if @options[:verbose]
+          next
+        end
+
+        if href.scheme and href.scheme != 'http' and href.scheme != 'https'
+          puts "Skipping append_query_string for: #{href.to_s} because scheme is no good" if @options[:verbose]
+          next
+        end
+
+        if href.query and not href.query.empty?
+          href.query = href.query + '&amp;' + qs
+        else
+          href.query = qs
+        end
+
+        el['href'] = href.to_s
+      rescue URI::Error => e
+        $stderr.puts "Skipping append_query_string for: #{href.to_s} (#{e.message})" if @options[:verbose]
+        next
+      end
+
+    end
+    doc
+  end
+
+  # Check for an XHTML doctype
+  def is_xhtml?
+    intro = @doc.to_html.strip.split("\n")[0..2].join(' ')
+    is_xhtml = !!(intro =~ /w3c\/\/[\s]*dtd[\s]+xhtml/i)
+    $stderr.puts "Is XHTML? #{is_xhtml.inspect}\nChecked:\n#{intro}" if @options[:debug]
+    is_xhtml
+  end
+
+  # Convert relative links to absolute links.
+  #
+  # Processes <tt>href</tt> <tt>src</tt> and <tt>background</tt> attributes
+  # as well as CSS <tt>url()</tt> declarations found in inline <tt>style</tt> attributes.
+  #
+  # <tt>doc</tt> is an Hpricot document and <tt>base_uri</tt> is either a string or a URI.
+  #
+  # Returns an Hpricot document.
+  def convert_inline_links(doc, base_uri) # :nodoc:
+    base_uri = URI.parse(base_uri) unless base_uri.kind_of?(URI)
+
+    append_qs = @options[:link_query_string] || ''
+
+    ['href', 'src', 'background'].each do |attribute|
+      tags = doc.search("*[@#{attribute}]")
+
+      next if tags.empty?
+
+      tags.each do |tag|
+        # skip links that look like they have merge tags
+        # and mailto, ftp, etc...
+        if tag.attributes[attribute].to_s =~ /^([\%\<\{\#\[]|data:|tel:|file:|sms:|callto:|facetime:|mailto:|ftp:|gopher:|cid:)/i
+          next
+        end
+
+        if tag.attributes[attribute].to_s =~ /^http/i
+          begin
+            merged = URI.parse(tag.attributes[attribute])
+          rescue; next; end
+        else
+          begin
+            merged = Premailer.resolve_link(tag.attributes[attribute].to_s, base_uri)
+          rescue
+            begin
+              merged = Premailer.resolve_link(URI.escape(tag.attributes[attribute].to_s), base_uri)
+            rescue; end
+          end
+        end
+
+        # make sure 'merged' is a URI
+        merged = URI.parse(merged.to_s) unless merged.kind_of?(URI)
+        tag[attribute] = merged.to_s
+      end # end of each tag
+    end # end of each attrs
+
+    doc.search("*[@style]").each do |el|
+      el['style'] = CssParser.convert_uris(el.attributes['style'].to_s, base_uri)
+    end
+    doc
+  end
+
+
+  # @private
+  def self.escape_string(str) # :nodoc:
+    str.gsub(/"/ , "'")
+  end
+
+  # @private
+  def self.resolve_link(path, base_path) # :nodoc:
+    path.strip!
+    resolved = nil
+    if path =~ /(http[s]?|ftp):\/\//i
+      resolved = path
+      Premailer.canonicalize(resolved)
+    elsif base_path.kind_of?(URI)
+      resolved = base_path.merge(path)
+      Premailer.canonicalize(resolved)
+    elsif base_path.kind_of?(String) and base_path =~ /\A(http[s]?|ftp):\/\//i
+      resolved = URI.parse(base_path)
+      resolved = resolved.merge(path)
+      Premailer.canonicalize(resolved)
+    else
+      File.expand_path(path, File.dirname(base_path))
+    end
+  end
+
+  # Test the passed variable to see if we are in local or remote mode.
+  #
+  # IO objects return true, as do strings that look like URLs.
+  def self.local_data?(data)
+    return true if data.is_a?(IO) || data.is_a?(StringIO)
+    return false if data =~ /\A(http|https|ftp)\:\/\//i
+    true
+  end
+
+  # from http://www.ruby-forum.com/topic/140101
+  def self.canonicalize(uri) # :nodoc:
+    u = uri.kind_of?(URI) ? uri : URI.parse(uri.to_s)
+    u.normalize!
+    newpath = u.path
+    while newpath.gsub!(%r{([^/]+)/\.\./?}) { |match|
+        $1 == '..' ? match : ''
+      } do end
+      newpath = newpath.gsub(%r{/\./}, '/').sub(%r{/\.\z}, '/')
+      u.path = newpath
+      u.to_s
+    end
+
+  # Check <tt>CLIENT_SUPPORT_FILE</tt> for any CSS warnings
+  def check_client_support # :nodoc:
+    @client_support ||= YAML::load(File.open(CLIENT_SUPPORT_FILE))
+
+    warnings = []
+    properties = []
+
+    # Get a list off CSS properties
+    @processed_doc.search("*[@style]").each do |el|
+      style_url = el.attributes['style'].to_s.gsub(/([\w\-]+)[\s]*\:/i) do |s|
+        properties.push($1)
+      end
+    end
+
+    properties.uniq!
+
+    property_support = @client_support['css_properties']
+    properties.each do |prop|
+      if property_support.include?(prop) and
+          property_support[prop].include?('support') and
+          property_support[prop]['support'] >= @options[:warn_level]
+        warnings.push({:message => "#{prop} CSS property",
+            :level => WARN_LABEL[property_support[prop]['support']],
+            :clients => property_support[prop]['unsupported_in'].join(', ')})
+      end
+    end
+
+    @client_support['attributes'].each do |attribute, data|
+      next unless data['support'] >= @options[:warn_level]
+      if @doc.search("*[@#{attribute}]").length > 0
+        warnings.push({:message => "#{attribute} HTML attribute",
+            :level => WARN_LABEL[data['support']],
+            :clients => data['unsupported_in'].join(', ')})
+      end
+    end
+
+    @client_support['elements'].each do |element, data|
+      next unless data['support'] >= @options[:warn_level]
+      if @doc.search(element).length > 0
+        warnings.push({:message => "#{element} HTML element",
+            :level => WARN_LABEL[data['support']],
+            :clients => data['unsupported_in'].join(', ')})
+      end
+    end
+
+    return warnings
+  end
+end
+