Podcast XML/Builder Template for Middleman

config.rb

page '/podcast.xml', layout: false

# Methods defined in the helpers block are available in templates
helpers do
  def podcast_source_path(article)
    "source/audio/#{article.data.file}"
  end
end

podcast-number-one.html.markdown

title	date	file
Podcast Number One	2014-12-12 12:40 CST	one.mp3

Blah Blah.

podcast.xml.builder

# -*- coding: utf-8 -*-
require 'mp3info'
require 'nokogiri'

# The main method on this module +plain_text+ will convert a string of HTML to a plain text approximation.
class HtmlToPlainText

  IGNORE_TAGS = %w(script style object applet iframe).inject({}){|h, t| h[t] = true; h}.freeze
  PARAGRAPH_TAGS = %w(p h1 h2 h3 h4 h5 h6 table ol ul dl dd blockquote dialog figure aside section).inject({}){|h, t| h[t] = true; h}.freeze
  BLOCK_TAGS = %w(div address li dt center del article header header footer nav pre legend tr).inject({}){|h, t| h[t] = true; h}.freeze
  WHITESPACE = [" ", "\n", "\r"].freeze
  PLAINTEXT = "plaintext".freeze
  PRE = "pre".freeze
  BR = "br".freeze
  HR = "hr".freeze
  TD = "td".freeze
  TH = "th".freeze
  TR = "tr".freeze
  OL = "ol".freeze
  UL = "ul".freeze
  LI = "li".freeze
  A = "a".freeze
  TABLE = "table".freeze
  NUMBERS = ["1", "a"].freeze
  ABSOLUTE_URL_PATTERN = /^[a-z]+:\/\/[a-z0-9]/i.freeze
  HTML_PATTERN = /[<&]/.freeze
  TRAILING_WHITESPACE = /[ \t]+$/.freeze
  BODY_TAG_XPATH = "/html/body".freeze
  CARRIDGE_RETURN_PATTERN = /\r(\n?)/.freeze
  LINE_BREAK_PATTERN = /[\n\r]/.freeze
  NON_PROTOCOL_PATTERN = /:\/?\/?(.*)/.freeze
  NOT_WHITESPACE_PATTERN = /\S/.freeze
  SPACE = " ".freeze
  EMPTY = "".freeze
  NEWLINE = "\n".freeze
  HREF = "href".freeze
  TABLE_SEPARATOR = " | ".freeze

  class << self
    # Convert some HTML into a plain text approximation.

    def truncate(str, truncate_at, options = {})
      return str.dup unless str.length > truncate_at

      options[:omission] ||= '...'
      length_with_room_for_omission = truncate_at - options[:omission].length
      stop =        if options[:separator]
        str.rindex(options[:separator], length_with_room_for_omission) || length_with_room_for_omission
      else
        length_with_room_for_omission
      end

      "#{str[0...stop]}#{options[:omission]}"
    end

    def plain_text(html)
      return nil if html.nil?
      return html.dup unless html =~ HTML_PATTERN
      body = Nokogiri::HTML::Document.parse(html).xpath(BODY_TAG_XPATH).first
      return unless body
      convert_node_to_plain_text(body).strip.gsub(CARRIDGE_RETURN_PATTERN, NEWLINE)
    end

    private

    # Convert an HTML node to plain text. This method is called recursively with the output and
    # formatting options for special tags.
    def convert_node_to_plain_text(parent, out = '', options = {})
      if PARAGRAPH_TAGS.include?(parent.name)
        append_paragraph_breaks(out)
      elsif BLOCK_TAGS.include?(parent.name)
        append_block_breaks(out)
      end

      format_list_item(out, options) if parent.name == LI
      out << "| " if parent.name == TR && data_table?(parent.parent)

      parent.children.each do |node|
        if node.text? || node.cdata?
          text = node.text
          unless options[:pre]
            text = node.text.gsub(LINE_BREAK_PATTERN, SPACE).squeeze(SPACE)
            text.lstrip! if WHITESPACE.include?(out[-1, 1])
          end
          out << text
        elsif node.name == PLAINTEXT
          out << node.text
        elsif node.element? && !IGNORE_TAGS.include?(node.name)
          convert_node_to_plain_text(node, out, child_options(node, options))

          if node.name == BR
            out.sub!(TRAILING_WHITESPACE, EMPTY)
            out << NEWLINE
          elsif node.name == HR
            out.sub!(TRAILING_WHITESPACE, EMPTY)
            out << NEWLINE unless out.end_with?(NEWLINE)
            out << "-------------------------------\n"
          elsif node.name == TD || node.name == TH
            out << (data_table?(parent.parent) ? TABLE_SEPARATOR : SPACE)
          elsif node.name == A
            href = node[HREF]
            if href &&
                href =~ ABSOLUTE_URL_PATTERN &&
                node.text =~ NOT_WHITESPACE_PATTERN &&
                node.text != href &&
                node.text != href[NON_PROTOCOL_PATTERN, 1] # use only text for <a href="mailto:a@b.com">a@b.com</a>
              out << " (#{href}) "
            end
          elsif PARAGRAPH_TAGS.include?(node.name)
            append_paragraph_breaks(out)
          elsif BLOCK_TAGS.include?(node.name)
            append_block_breaks(out)
          end
        end
      end
      out
    end

    # Set formatting options that will be passed to child elements for a tag.
    def child_options(node, options)
      if node.name == UL
        level = options[:ul] || -1
        level += 1
        options.merge(:list => :ul, :ul => level)
      elsif node.name == OL
        level = options[:ol] || -1
        level += 1
        options.merge(:list => :ol, :ol => level, :number => NUMBERS[level % 2])
      elsif node.name == PRE
        options.merge(:pre => true)
      else
        options
      end
    end

    # Add double line breaks between paragraph elements. If line breaks already exist,
    # new ones will only be added to get to two.
    def append_paragraph_breaks(out)
      out.sub!(TRAILING_WHITESPACE, EMPTY)
      if out.end_with?(NEWLINE)
        out << NEWLINE unless out.end_with?("\n\n")
      else
        out << "\n\n"
      end
    end

    # Add a single line break between block elements. If a line break already exists,
    # none will be added.
    def append_block_breaks(out)
      out.sub!(TRAILING_WHITESPACE, EMPTY)
      out << NEWLINE unless out.end_with?(NEWLINE)
    end

    # Add an appropriate bullet or number to a list element.
    def format_list_item(out, options)
      if options[:list] == :ul
        out << "#{'*' * (options[:ul] + 1)} "
      elsif options[:list] == :ol
        number = options[:number]
        options[:number] = number.next
        out << "#{number}. "
      end
    end

    def data_table?(table)
      table.attributes['border'].to_s.to_i > 0
    end
  end
end

xml.instruct!
xml.rss 'xmlns:itunes' => 'http://www.itunes.com/dtds/podcast-1.0.dtd', 'xmlns:atom' => 'http://www.w3.org/2005/Atom', :version => '2.0' do
  xml.channel do
    xml.title config[:blog_title]
    xml.description config[:blog_description]
    xml.link URI.join(config[:blog_url], blog.options.prefix.to_s, 'podcast.xml')
    xml.atom :link, 'rel' => 'self', 'href' => URI.join(config[:blog_url], blog.options.prefix.to_s, 'podcast.xml')
    xml.language 'en-CA'
    xml.lastBuildDate blog.articles.first.date.rfc2822
    xml.pubDate blog.articles.first.date.rfc2822
    xml.itunes :author, 'John Frank'
    xml.itunes :keywords, config[:blog_keywords].join(', ')
    xml.itunes :explicit, (config[:blog_clean] ? 'clean' : 'yes')
    xml.itunes :image, :href => URI.join(config[:blog_url], image_path('icon.png'))
    xml.itunes :summary, HtmlToPlainText.truncate(HtmlToPlainText.plain_text(config[:blog_description]), 1950)
    xml.itunes :owner do
      xml.itunes :name, 'John Frank'
      xml.itunes :email, 'john@example.com'
    end
    xml.itunes :category, :text => 'Science & Medicine' do
      xml.itunes :category, :text => 'Medicine'
    end

    blog.articles.each do |article|
      xml.item do
        xml.title article.title
        xml.pubDate article.date.rfc822
        xml.enclosure :url => tracked_url(podcast_url(article)), :length => File.size(podcast_source_path(article)), :type => 'audio/mpeg'
        xml.link URI.join(config[:blog_url], article.url)
        xml.guid({ :isPermaLink => true }, URI.join(config[:blog_url], article.url))
        xml.itunes :author, 'John Frank'
        xml.itunes :summary do
          xml.cdata! HtmlToPlainText.truncate(HtmlToPlainText.plain_text(article.body), 3950)
        end
        xml.itunes :duration, Mp3Info.new(podcast_source_path(article)).length.to_i
        # xml.description do
        #   xml.cdata! article.body + partial(:audio_tag, :locals => { :article => article })
        # end
        # Most RSS readers will pull out the link to the enclosure, so no need to include it here.
        xml.description do
          xml.cdata! article.body
        end
      end
    end
  end
end