gistfile1.txt

#nokogiri callbacks
require_relative 'wiki_handler'

class WikiImport < Nokogiri::XML::SAX::Document
  
  include WikiHandler
  
  # A stack (Array) of attributes as we find them 
  attr_accessor :attribute_stack
  
  # A logger to output to the screen
  attr_accessor :logger
  
  # A counter to increment each time you find a page
  attr_accessor :page_count
  
  # The output SQL file
  attr_accessor :sql
  
  # The contents of the last page as a hash
  attr_accessor :last_page
  
  # The text contents of last element's body 
  attr_accessor :last_body

  attr_accessor :sql
  
  def initialize(logger)
    self.logger = logger
    self.attribute_stack = Array.new
    self.page_count = 0
    self.last_page = {}
    self.last_body = ""
    @output_file_count = 0
    @title = ""
    @text = ""
  end
  
  def start_document

  end
  
  def end_document
    # logger.debug "End document"
  end

  def start_element(name, attrs)
    # logger.debug "Found element #{name}"
    
    case name
    when "title"
      @interestedtitle = true
    
    when "text"
      @interestedtext = true
    else 
      @interestedtitle = false
      @interestedtext = false
    end
    
    # logger.debug "Found element #{name}"
  end

  def characters(c)
    @title = @title + c if @interestedtitle 
    #@title += c if interestedtitle
    @text = @text + c if @interestedtext
  end
  
  def end_element(name)
    #say we have finished the page and then print out the title and body
    if name == 'page'
      sql = File.open('data/my_stuff.sql', 'w')
      sql << "INSERT INTO articles(title, body) VALUES('#{@title}', '#{@text}');"
      sql.close
    puts @title
    puts @text
    # logger.debug @text

      @title = "" 
      @text = "" 
    end

  end
  
  def method_missing(m, *args, &block)
    # logger.debug("Ignoring #{m}")
  end
  
  protected
  def handler_method(name)
    :"handle_#{name.downcase}"
  end
  
  def clean(s)
    s.strip.gsub("'", "''")
    #this has the effect of taking out the apostrohies
  end
  
  def output_file_name
    "/tmp/articles-#{@output_file_count}.sql"
  end  

  # results []

  # page = doc.css('page').text
  # page = doc.css('title').text
  # page = doc.css('body').text

  # results << page
  # results << title
  # results << body

  # binindg.pry


end