#!/usr/bin/env ruby
###
# parse_lumpley.rb
#
# A script to read Vincent Baker's weblog and make an RSS feed from it.
#
###

require 'net/http'
require 'cgi'
require 'rexml/document'
include REXML

class Post
   attr_reader :id, :body, :date
   
   def initialize(id, body, date)
      @id = id
      @body = body
      @date = date
   end
end

class Feed
   def initialize(title, link, copyright, generator, comment_link)
      @posts = []
      @title = title
      @link = link
      @copyright = copyright
      @generator = generator
      @pubDate = Time.now
      @comment_link = comment_link
   end
   
   def add(id, body, date)
      @posts[@posts.length] = Post.new(id, body, date)
   end
   
   def to_s
      xml = Document.new('<rss version="2.0"></rss>')
      channel = Element.new('channel')
      channel.add_element(Element.new('title').add_text(@title))
      channel.add_element(Element.new('link').add_text(@link))
      channel.add_element(Element.new('copyright').add_text(@copyright))
      channel.add_element(Element.new('generator').add_text(@generator))
      channel.add_element(Element.new('pubDate').add_text(@pubDate.to_s))
      xml.root[0,0] = channel
      
      @posts.each do |post|
          item = Element.new('item')
          item.add_element('pubDate').add_text(post.date.to_s)
          item.add_element('link').add_text(@link + "#" + post.id)
          item.add_element('guid').add_text(@link + "#" + post.id)
          item.add_element('comment').add_text(@comment_link + post.id)
          item.add_element('description').add_text(post.body.gsub(%r{<[^>]+>}, '')[0..150] + "...")
          item.add_element('content:encoded').add_text(post.body)
          channel.add_element(item)
      end
      
      xml
   end
end

response = Net::HTTP.get_response('www.lumpley.com', '/opine.html')

feed = Feed.new('Anyway - Lumpley\'s Weblog', 'http://www.lumpley.com/opine.html', 'Copyright 2005 Vincent Baker', 'lumpley_rss.rb', 'http://www.lumpley.com/anycomment.php?entry=')
post_num = nil
post_text = ''
post_date = nil
start = false

response.body.each do |line|
   if line =~ /<!-- newnews -->/ .. line =~ /<!-- endnews -->/
      if line =~ /<!-- (\d+) -->/
         post_num = $1
         post_text = ''
         post_date = nil
      elsif line =~ /<!-- (com|end)#{post_num} -->/
         print post_date
         feed.add(post_num, post_text, post_date)
         post_num = nil
      elsif !(post_num.nil?)
         if line =~ /<\/p>\s+$/
            start = false
         elsif line =~ /^<p>/
            start = true
            if line =~ %r{^<p><b>(\d+)-(\d+)-(\d+)</b><br />}
               post_date = Time.utc(2000 + $3.to_i, $1.to_i, $2.to_i)         
            end
         elsif start == true
            post_text += line
         end
      end
   end
end

puts "HTTP/1.0 200 OK\r\n"
puts "Content-type: text/xml\r\n\r\n"
puts feed.to_s