module RSS::ListenerMixin

Constants

AVAILABLE_PARSERS

The list of all available parsers, in constant form.

AVAILABLE_PARSER_LIBRARIES

The list of all available libraries for parsing.

CONTENT_PATTERN
NAMESPLIT

Attributes

do_validate[RW]
ignore_unknown_element[RW]
rss[R]

Public Class Methods

new() click to toggle source
# File lib/rss/parser.rb, line 268
def initialize
  @rss = nil
  @ignore_unknown_element = true
  @do_validate = true
  @ns_stack = [{"xml" => :xml}]
  @tag_stack = [[]]
  @text_stack = ['']
  @proc_stack = []
  @last_element = nil
  @version = @encoding = @standalone = nil
  @xml_stylesheets = []
  @xml_child_mode = false
  @xml_element = nil
  @last_xml_element = nil
end

Public Instance Methods

instruction(name, content) click to toggle source
# File lib/rss/parser.rb, line 289
def instruction(name, content)
  if name == "xml-stylesheet"
    params = parse_pi_content(content)
    if params.has_key?("href")
      @xml_stylesheets << XMLStyleSheet.new(params)
    end
  end
end
tag_end(name) click to toggle source
# File lib/rss/parser.rb, line 351
def tag_end(name)
  if DEBUG
    p "end tag #{name}"
    p @tag_stack
  end
  text = @text_stack.pop
  tags = @tag_stack.pop
  pr = @proc_stack.pop
  pr.call(text, tags) unless pr.nil?
  @ns_stack.pop
end
tag_start(name, attributes) click to toggle source
# File lib/rss/parser.rb, line 298
def tag_start(name, attributes)
  @text_stack.push('')

  ns = @ns_stack.last.dup
  attrs = {}
  attributes.each do |n, v|
    if /\Axmlns(?:\z|:)/ =~ n
      ns[$POSTMATCH] = v
    else
      attrs[n] = v
    end
  end
  @ns_stack.push(ns)

  prefix, local = split_name(name)
  @tag_stack.last.push([_ns(ns, prefix), local])
  @tag_stack.push([])
  if @xml_child_mode
    previous = @last_xml_element
    element_attrs = attributes.dup
    unless previous
      ns.each do |ns_prefix, value|
        next if ns_prefix == "xml"
        key = ns_prefix.empty? ? "xmlns" : "xmlns:#{ns_prefix}"
        element_attrs[key] ||= value
      end
    end
    next_element = XML::Element.new(local,
                                    prefix.empty? ? nil : prefix,
                                    _ns(ns, prefix),
                                    element_attrs)
    previous << next_element if previous
    @last_xml_element = next_element
    pr = Proc.new do |text, tags|
      if previous
        @last_xml_element = previous
      else
        @xml_element = @last_xml_element
        @last_xml_element = nil
      end
    end
    @proc_stack.push(pr)
  else
    if @rss.nil? and respond_to?("initial_start_#{local}", true)
      __send__("initial_start_#{local}", local, prefix, attrs, ns.dup)
    elsif respond_to?("start_#{local}", true)
      __send__("start_#{local}", local, prefix, attrs, ns.dup)
    else
      start_else_element(local, prefix, attrs, ns.dup)
    end
  end
end
text(data) click to toggle source
# File lib/rss/parser.rb, line 363
def text(data)
  if @xml_child_mode
    @last_xml_element << data if @last_xml_element
  else
    @text_stack.last << data
  end
end
xmldecl(version, encoding, standalone) click to toggle source

set instance vars for version, encoding, standalone

# File lib/rss/parser.rb, line 285
def xmldecl(version, encoding, standalone)
  @version, @encoding, @standalone = version, encoding, standalone
end

Private Instance Methods

_ns(ns, prefix) click to toggle source
# File lib/rss/parser.rb, line 372
def _ns(ns, prefix)
  ns.fetch(prefix, "")
end
check_ns(tag_name, prefix, ns, require_uri, ignore_unknown_element=nil) click to toggle source
# File lib/rss/parser.rb, line 427
def check_ns(tag_name, prefix, ns, require_uri, ignore_unknown_element=nil)
  if _ns(ns, prefix) == require_uri
    true
  else
    if ignore_unknown_element.nil?
      ignore_unknown_element = @ignore_unknown_element
    end

    if ignore_unknown_element
      false
    elsif @do_validate
      raise NSError.new(tag_name, prefix, require_uri)
    else
      # Force bind required URI with prefix
      @ns_stack.last[prefix] = require_uri
      true
    end
  end
end
collect_attributes(tag_name, prefix, attrs, ns, klass) click to toggle source
# File lib/rss/parser.rb, line 477
  def collect_attributes(tag_name, prefix, attrs, ns, klass)
    attributes = {}
    klass.get_attributes.each do |a_name, a_uri, required, element_name|
      if a_uri.is_a?(String) or !a_uri.respond_to?(:include?)
        a_uri = [a_uri]
      end
      unless a_uri == [""]
        for prefix, uri in ns
          if a_uri.include?(uri)
            val = attrs["#{prefix}:#{a_name}"]
            break if val
          end
        end
      end
      if val.nil? and a_uri.include?("")
        val = attrs[a_name]
      end

      if @do_validate and required and val.nil?
        unless a_uri.include?("")
          for prefix, uri in ns
            if a_uri.include?(uri)
              a_name = "#{prefix}:#{a_name}"
            end
          end
        end
        raise MissingAttributeError.new(tag_name, a_name)
      end

      attributes[a_name] = val
    end
    attributes
  end

  def setup_next_element(tag_name, klass, attributes)
    previous = @last_element
    next_element = klass.new(@do_validate, attributes)
    previous.set_next_element(tag_name, next_element)
    @last_element = next_element
    @last_element.parent = previous if klass.need_parent?
    @xml_child_mode = @last_element.have_xml_content?

    Proc.new do |text, tags|
      p(@last_element.class) if DEBUG
      if @xml_child_mode
        @last_element.content = @xml_element.to_s
        xml_setter = @last_element.class.xml_setter
        @last_element.__send__(xml_setter, @xml_element)
        @xml_element = nil
        @xml_child_mode = false
      else
        if klass.have_content?
          if @last_element.need_base64_encode?
            text = text.lstrip.unpack("m").first
          end
          @last_element.content = text
        end
      end
      if @do_validate
        @last_element.validate_for_stream(tags, @ignore_unknown_element)
      end
      @last_element = previous
    end
  end

  def setup_next_element_in_unknown_element
    current_element, @last_element = @last_element, nil
    Proc.new {@last_element = current_element}
  end
end
initial_start_RDF(tag_name, prefix, attrs, ns) click to toggle source
# File lib/rss/1.0.rb, line 471
def initial_start_RDF(tag_name, prefix, attrs, ns)
  check_ns(tag_name, prefix, ns, RDF::URI, false)

  @rss = RDF.new(@version, @encoding, @standalone)
  @rss.do_validate = @do_validate
  @rss.xml_stylesheets = @xml_stylesheets
  @last_element = @rss
  pr = Proc.new do |text, tags|
    @rss.validate_for_stream(tags, @ignore_unknown_element) if @do_validate
  end
  @proc_stack.push(pr)
end
initial_start_entry(tag_name, prefix, attrs, ns) click to toggle source
# File lib/rss/atom.rb, line 1010
def initial_start_entry(tag_name, prefix, attrs, ns)
  check_ns(tag_name, prefix, ns, Atom::URI, false)

  @rss = Atom::Entry.new(@version, @encoding, @standalone)
  @rss.do_validate = @do_validate
  @rss.xml_stylesheets = @xml_stylesheets
  @rss.lang = attrs["xml:lang"]
  @rss.base = attrs["xml:base"]
  @last_element = @rss
  pr = Proc.new do |text, tags|
    @rss.validate_for_stream(tags) if @do_validate
  end
  @proc_stack.push(pr)
end
initial_start_feed(tag_name, prefix, attrs, ns) click to toggle source
# File lib/rss/atom.rb, line 995
def initial_start_feed(tag_name, prefix, attrs, ns)
  check_ns(tag_name, prefix, ns, Atom::URI, false)

  @rss = Atom::Feed.new(@version, @encoding, @standalone)
  @rss.do_validate = @do_validate
  @rss.xml_stylesheets = @xml_stylesheets
  @rss.lang = attrs["xml:lang"]
  @rss.base = attrs["xml:base"]
  @last_element = @rss
  pr = Proc.new do |text, tags|
    @rss.validate_for_stream(tags) if @do_validate
  end
  @proc_stack.push(pr)
end
initial_start_rss(tag_name, prefix, attrs, ns) click to toggle source
# File lib/rss/0.9.rb, line 447
def initial_start_rss(tag_name, prefix, attrs, ns)
  check_ns(tag_name, prefix, ns, "", false)

  @rss = Rss.new(attrs['version'], @version, @encoding, @standalone)
  @rss.do_validate = @do_validate
  @rss.xml_stylesheets = @xml_stylesheets
  @last_element = @rss
  pr = Proc.new do |text, tags|
    @rss.validate_for_stream(tags, @ignore_unknown_element) if @do_validate
  end
  @proc_stack.push(pr)
end
known_class?(target_class, class_name) click to toggle source
# File lib/rss/parser.rb, line 408
def known_class?(target_class, class_name)
  class_name and
    (target_class.const_defined?(class_name, false) or
     target_class.constants.include?(class_name.to_sym))
end
parse_pi_content(content) click to toggle source

Extract the first name=“value” pair from content. Works with single quotes according to the constant CONTENT_PATTERN. Return a Hash.

# File lib/rss/parser.rb, line 380
def parse_pi_content(content)
  params = {}
  content.scan(CONTENT_PATTERN) do |name, quote, value|
    params[name] = value
  end
  params
end
setup_next_element(tag_name, klass, attributes) click to toggle source
# File lib/rss/parser.rb, line 511
def setup_next_element(tag_name, klass, attributes)
  previous = @last_element
  next_element = klass.new(@do_validate, attributes)
  previous.set_next_element(tag_name, next_element)
  @last_element = next_element
  @last_element.parent = previous if klass.need_parent?
  @xml_child_mode = @last_element.have_xml_content?

  Proc.new do |text, tags|
    p(@last_element.class) if DEBUG
    if @xml_child_mode
      @last_element.content = @xml_element.to_s
      xml_setter = @last_element.class.xml_setter
      @last_element.__send__(xml_setter, @xml_element)
      @xml_element = nil
      @xml_child_mode = false
    else
      if klass.have_content?
        if @last_element.need_base64_encode?
          text = text.lstrip.unpack("m").first
        end
        @last_element.content = text
      end
    end
    if @do_validate
      @last_element.validate_for_stream(tags, @ignore_unknown_element)
    end
    @last_element = previous
  end
end
setup_next_element_in_unknown_element() click to toggle source
# File lib/rss/parser.rb, line 542
def setup_next_element_in_unknown_element
  current_element, @last_element = @last_element, nil
  Proc.new {@last_element = current_element}
end
split_name(name) click to toggle source
# File lib/rss/parser.rb, line 422
def split_name(name)
  name =~ NAMESPLIT
  [$1 || '', $2]
end
start_else_element(local, prefix, attrs, ns) click to toggle source
# File lib/rss/parser.rb, line 388
def start_else_element(local, prefix, attrs, ns)
  class_name = self.class.class_name(_ns(ns, prefix), local)
  current_class = @last_element.class
  if known_class?(current_class, class_name)
    next_class = current_class.const_get(class_name)
    start_have_something_element(local, prefix, attrs, ns, next_class)
  else
    if !@do_validate or @ignore_unknown_element
      @proc_stack.push(setup_next_element_in_unknown_element)
    else
      parent = "ROOT ELEMENT???"
      if current_class.tag_name
        parent = current_class.tag_name
      end
      raise NotExpectedTagError.new(local, _ns(ns, prefix), parent)
    end
  end
end
start_get_text_element(tag_name, prefix, ns, required_uri) click to toggle source
# File lib/rss/parser.rb, line 447
def start_get_text_element(tag_name, prefix, ns, required_uri)
  pr = Proc.new do |text, tags|
    setter = self.class.setter(required_uri, tag_name)
    if setter and @last_element.respond_to?(setter)
      if @do_validate
        getter = self.class.getter(required_uri, tag_name)
        if @last_element.__send__(getter)
          raise TooMuchTagError.new(tag_name, @last_element.tag_name)
        end
      end
      @last_element.__send__(setter, text.to_s)
    else
      if @do_validate and !@ignore_unknown_element
        raise NotExpectedTagError.new(tag_name, _ns(ns, prefix),
                                      @last_element.tag_name)
      end
    end
  end
  @proc_stack.push(pr)
end
start_have_something_element(tag_name, prefix, attrs, ns, klass) click to toggle source
# File lib/rss/parser.rb, line 468
def start_have_something_element(tag_name, prefix, attrs, ns, klass)
  if check_ns(tag_name, prefix, ns, klass.required_uri)
    attributes = collect_attributes(tag_name, prefix, attrs, ns, klass)
    @proc_stack.push(setup_next_element(tag_name, klass, attributes))
  else
    @proc_stack.push(setup_next_element_in_unknown_element)
  end
end