require "rubygems"
require "hpricot"

# Apricot eats Gorilla is a SOAP communication helper.
# 
# It translates between SOAP messages (XML) and Ruby Hashes while offering some
# helpful methods for working with SOAP webservices. Apricot eats Gorilla was
# initially based on CobraVsMongoose but uses Hpricot instead of REXML.
class ApricotEatsGorilla
  class << self # Class methods

    # Flag to enable sorting of Hash keys.
    attr_accessor :sort_keys

    # Flag to disable conversion of tag names to lowerCamelCase.
    attr_accessor :disable_tag_names_to_lower_camel_case

    # Flag to disable conversion of Hash keys to snake_case.
    attr_accessor :disable_hash_keys_to_snake_case

    # Flag to disable conversion of Hash keys to Symbols.
    attr_accessor :disable_hash_keys_to_symbol

    # Array of XML nodes to add a namespace to.
    attr_accessor :nodes_to_namespace

    # The namespace for nodes set by nodes_to_namespace.
    attr_accessor :node_namespace

    # Shortcut method for translating between XML Strings and Ruby Hashes.
    # Delegates to xml_to_hash in case +source+ is of type String or delegates
    # to hash_to_xml in case +source+ is of type Hash. Returns nil otherwise.
    #
    # ==== Parameters
    #
    # * +source+ - The XML String or Ruby Hash to translate.
    # * +root_node+ - Optional. Custom root node to start parsing the given XML.
    def [](source, root_node = nil)
      case source
      when String
        xml_to_hash(source, root_node)
      when Hash
        hash_to_xml(source)
      else
        nil
      end
    end

    # Yields this class object in case a +block+ was given. Nice way for setting
    # multiple options at once.
    def setup
      yield self if block_given?
    end

    # Converts a given +xml+ String into a Ruby Hash. Starts parsing at root
    # node by default. The optional +root_node+ parameter can be used to specify
    # a custom root node to start parsing at using an XPath (Hpricot search).
    # The root node itself won't be included in the Hash. Converts tag names
    # from CamelCase/lowerCamelCase to snake_case and into Symbols by default.
    #
    # ==== Parameters
    #
    # * +xml+ - The XML String to translate.
    # * +root_node+ - Optional. Custom root node to start parsing the given XML.
    #
    # ==== Examples
    #
    #   xml = "<apricot><eats>Gorilla</eats></apricot>"
    #   ApricotEatsGorilla[xml]
    #   # => { :eats => "Gorilla" }
    #
    #   xml = "<apricot><eats><lotsOf>Gorillas</lotsOf></eats></apricot>"
    #   ApricotEatsGorilla[xml, "//eats"]
    #   # => { :lots_of => "Gorillas" }
    def xml_to_hash(xml, root_node = nil)
      doc = Hpricot.XML clean_xml(xml)
      root = root_node ? doc.search(root_node) : doc.root

      return nil if root.nil?
      # root is an Hpricot::Elements object
      if root.respond_to?(:count)
        # root is an Hpricot::Elements object with only one element
        return root.entries.map { |e| xml_node_to_hash(e) } if root.count > 1
        root = root.entries.first
        if root.children.size == 1 && root.children.first.text?
          # root contains text only
          return root.children.first.to_s
        end
      end
      xml_node_to_hash(root)
    end

    # Converts a given Ruby Hash into an XML String. Converts Hash keys from
    # snake_case to lowerCamelCase by default.
    #
    # ==== Parameters
    #
    # * +hash+ - The Ruby Hash to translate.
    #
    # ==== Examples
    #
    #   hash = { :apricot => { :eats => "Gorilla" } }
    #   ApricotEatsGorilla[hash]
    #   # => "<apricot><eats>Gorilla</eats></apricot>"
    #
    #   hash = { :apricot => { :eats => [ "Gorillas", "Snakes" ] } }
    #   ApricotEatsGorilla[hash]
    #   # => "<apricot><eats>Gorillas</eats><eats>Snakes</eats></apricot>"
    def hash_to_xml(hash)
      nested_data_to_xml(hash.keys.first, hash.values.first)
    end

    # Builds a SOAP request envelope and includes the content from a given +block+
    # into the envelope body. Accepts a Hash (namespace => namespace_uri) of
    # additional +namespaces+ to set.
    #
    # ==== Parameters
    #
    # * +namespaces+ - A Hash of namespaces and their URI.
    #
    # ==== Examples
    #
    #  ApricotEatsGorilla.soap_envelope { "<authenticate>me</authenticate>" }
    #
    #  # => '<env:Envelope xmlns:env="http://schemas.xmlsoap.org/soap/envelope/">
    #  # =>    <env:Body>
    #  # =>      <authenticate>me</authenticate>
    #  # =>    </env:Body>
    #  # =>  </env:Envelope>'
    #
    #  ApricotEatsGorilla.soap_envelope :wsdl => "http://example.com" do
    #    "<authenticate>me</authenticate>"
    #  end
    #
    #  # => '<env:Envelope
    #  # =>    xmlns:wsdl="http://example.com"
    #  # =>    xmlns:env="http://schemas.xmlsoap.org/soap/envelope/">
    #  # =>    <env:Body>
    #  # =>      <authenticate>me</authenticate>
    #  # =>    </env:Body>
    #  # =>  </env:Envelope>'
    def soap_envelope(namespaces = {})
      namespaces["env"] = "http://schemas.xmlsoap.org/soap/envelope/"
      tag("env:Envelope", namespaces) do
        tag("env:Body") do
          yield if block_given?
        end
      end
    end

    # Converts a given +string+ from CamelCase/lowerCamelCase to snake_case.
    def to_snake_case(string)
      string = string.gsub(/[A-Z]+/, '\1_\0').downcase
      string = string[1, string.length-1] if string[0, 1] == "_"
      string
    end

    # Converts a given +string+ from snake_case to lowerCamelCase.
    def to_lower_camel_case(string)
      string.to_s.gsub(/_(.)/) { $1.upcase }
    end

  private

    # Actual implementation for xml_to_hash. Takes and iterates through a given
    # Hpricot +element+ and returns a Ruby Hash equal to the given content.
    #
    # ==== Parameters
    #
    # * +element+ - The Hpricot element to translate.
    def xml_node_to_hash(element)
      this_node = {}
      element.each_child do |child|
        # hpricot 0.6.1 returns an empty array, while 0.8 returns nil
        if child.children.nil? || child.children.empty?
          key, value = child.name, nil
        elsif child.children.size == 1 && child.children.first.text?
          key, value = child.name, to_boolean(child.children.first.to_html)
        else
          key, value = child.name, xml_node_to_hash(child)
        end

        key = remove_namespace(key)
        key = to_snake_case(key) unless disable_hash_keys_to_snake_case
        key = key.intern unless disable_hash_keys_to_symbol
        current = this_node[key]
        case current
          when Array
            this_node[key] << value
          when nil
            this_node[key] = value
          else
            this_node[key] = [current.dup, value]
        end
      end
      this_node
    end

    # Actual implementation for hash_to_xml. Takes a Hash key +name+ and a
    # value +item+ and returns an XML String equal to the given content.
    #
    # ==== Parameters
    #
    # * +name+ - Root key from Hash to translate.
    # * +item+ - Root value from Hash to translate.
    def nested_data_to_xml(name, item)
      case item
      when Array
        item.map { |subitem| nested_data_to_xml(name, subitem) }.join
      when Hash
        tag(name) do
          opt_order(item).map { |tag, value|
            case value
            when Array
              value.map { |subitem| nested_data_to_xml(tag, subitem) }.join
            when Hash
              nested_data_to_xml(tag, value)
            else
              tag(tag) { value.to_s } if value.respond_to? :to_s
            end
          }.join
        end
      else
        tag(name) { item.to_s } if item.respond_to? :to_s
      end
    end

    # Creates an XML tag. Expects a block for tag content. Defaults to an empty
    # element tag in case no block was supplied.
    #
    # ==== Parameters
    #
    # * +name+ - The name of the XML tag.
    # * +attributes+ - Optional. Hash of attributes for the XML tag.
    def tag(name, attributes = {})
      name = to_lower_camel_case(name) unless disable_tag_names_to_lower_camel_case
      if nodes_to_namespace.kind_of? Array
        name = "#{node_namespace}:#{name}" if node_namespace && nodes_to_namespace.include?(name)
      end
      return "<#{name} />" unless block_given?

      attr = opt_order(attributes).map { |k, v| %Q( xmlns:#{k}="#{v}") }.to_s
      body = (yield && !yield.empty?) ? yield : ""
      "<#{name}#{attr}>" << body << "</#{name}>"
    end

    # Removes line breaks and whitespace between tags from a given +xml+ String.
    def clean_xml(xml)
      xml.gsub!(/\n+/, "")
      xml.gsub!(/(>)\s*(<)/, '\1\2')
      xml
    end

    # Removes the namespace from a given XML +tag+.
    def remove_namespace(tag)
      tag.sub!(/.+:(.+)/, '\1')
      tag
    end

    # Checks to see if a given +string+ matches "true" or "false" and converts
    # these values to actual Boolean objects. Returns the original String in
    # case it does not match "true" or "false".
    def to_boolean(string)
      return true if string == "true"
      return false if string == "false"
      string
    end

    # Returns a sorted version of a given +hash+ if sort_keys is enabled.
    def opt_order(hash)
      return hash unless sort_keys
      hash.keys.sort_by { |s| s.to_s }.map { |key| [key, hash[key]] }
    end

  end
end