Class: XOXO::Parser

Inherits:

Object

• Object
• XOXO::Parser

Defined in:

lib/xoxo.rb

Constant Summary

CONTAINER_TAGS =

%w{dl ol ul}

Instance Attribute Summary

• #structs ⇒ Array readonly

Instance Method Summary

• #initialize(xoxo) ⇒ Parser constructor

  Initialize new XOXO Parser.

• #normalize_attrs(attrs) ⇒ Hash private

  Take a hash of attributes and make sure the keys are all lowercase.

• #parse ⇒ Parser

  Parse XOXO document.

• #push(struct) ⇒ Object private

Constructor Details

#initialize(xoxo) ⇒ Parser

Initialize new XOXO Parser.

# File 'lib/xoxo.rb', line 165

def initialize(xoxo)
  @parser = REXML::Parsers::PullParser.new(xoxo)

  @textstack = ['']
  @xostack = []
  @structs = []
  @tags = []
end

Instance Attribute Details

#structs ⇒ Array (readonly)

Returns:

• (Array)

# File 'lib/xoxo.rb', line 162

def structs
  @structs
end

Instance Method Details

#normalize_attrs(attrs) ⇒ Hash (private)

Take a hash of attributes and make sure the keys are all lowercase.

Parameters:

• the (Hash) —

  attributes hash

Returns:

• (Hash) —

  the normalized attributes hash

# File 'lib/xoxo.rb', line 255

def normalize_attrs(attrs)
  attrs.keys.find_all { |k, v| k != k.downcase }.each { |k, v|
    v = v.downcase  if k == "rel" || k == "type"
    attrs.delete k
    attrs[k.downcase] = v
  }
  attrs
end

#parse ⇒ Parser

Parse XOXO document.

The end result of parsing is stored in the structs attribute.

Returns:

• (Parser) —

  the current parser object

# File 'lib/xoxo.rb', line 179

def parse
  while @parser.has_next?
    res = @parser.pull

    if res.start_element?
      @tags << res[0]

      case res[0]
      when "a"
        attrs = normalize_attrs res[1]
        attrs['url'] = attrs['href']
        attrs.delete 'href'
        push attrs
        @textstack << ''

      when "dl"
        push({})

      when "ol", "ul"
        push []

      when "li", "dt", "dd"
        @textstack << ''

      end
    elsif res.end_element?
      @tags.pop

      case res[0]
      when "a"
        val = @textstack.pop
        unless val.empty?
          val = ''  if @xostack.last['title'] == val
          val = ''  if @xostack.last['url'] == val
          @xostack.last['text'] = val  unless val.empty?
        end
        @xostack.pop

      when "dl", "ol", "ul"
        @xostack.pop

      when "li"
        val = @textstack.pop
        while @structs.last != @xostack.last
          val = @structs.pop
          @xostack.last << val
        end
        @xostack.last << val  if val.kind_of? String

      when "dt"
        # skip

      when "dd"
        val = @textstack.pop
        key = @textstack.pop

        val = @structs.pop  if @structs.last != @xostack.last
        @xostack.last[key] = val

      end
    elsif res.text?
      unless @tags.empty? || CONTAINER_TAGS.include?(@tags.last)
        @textstack.last << res[0]
      end
    end
  end

  self
end

#push(struct) ⇒ Object (private)

# File 'lib/xoxo.rb', line 265

def push(struct)
  if struct == {} && @structs.last.kind_of?(Hash) &&
      @structs.last.has_key?('url') &&
      @structs.last != @xostack.last
    # put back the <a>-made one for extra def's
    @xostack << @structs.last
  else
    @structs << struct
    @xostack << @structs.last
  end
end