#!/usr/bin/env ruby
#
# Author: IKARASHI Yoshinori <yoosee@gmail.com>
# $Id$
# License: GPLv2
#

require 'cgi'

require 'rss/parser'
require 'rss/0.9'
require 'rss/1.0'
require 'rss/2.0'
require 'rss/syndication'
require 'rss/dublincore'
require 'rss/content'

require 'uri'
require 'open-uri'
require 'timeout'

###########################
### Configuration parameters
BASEDIR = "/tmp/rss2ini/"

CACHETIME = 600
FETCHTIMEOUT = 30

OUTPUT_CHARSET = "Shift_JIS"
OUTPUT_FORMAT = "[%N%]
title=%TITLE%
link=%URI%
pubDate=%DATE%
description=%DESCRIPTION%

"
RSSURI = ''

###################

class Feed
  attr_accessor :uri, :output_charset, :output_format, :cachetime
  def initialize(uri, title = '')
    @basedir = BASEDIR
    @cachetime = CACHETIME
    @fetchtimeout = FETCHTIMEOUT
    @output_format = OUTPUT_FORMAT
    @output_charset = OUTPUT_CHARSET
    Dir.mkdir(@basedir) unless FileTest.directory?(@basedir)
    @uri = uri
    @title = title
    @rss = nil
  end

  def fetch
    rsssource = ''
    fetchfile = @basedir + @uri.gsub(/^https?:\/\//, '').gsub(/\//, '-').gsub(/-$/, '').to_s

    if(File.exist?(fetchfile) and 
       File.mtime(fetchfile) > Time.now - @cachetime)
      rssfile = File.new(fetchfile, 'r')
      rssfile.each_line do |line|
        rsssource << line
      end
      rssfile.close
    else
      rssfile = File.new(fetchfile, 'w')
      begin
        page = timeout(@fetchtimeout) { URI.parse(uri).read }
        rssfile.print page
        rsssource = page
      rescue OpenURI::HTTPError
        rsssource = nil
      rescue StandardError, TimeoutError
        rsssource = nil
      end
      rssfile.close
    end

    rsssource
  end

  def rss
    rsssource = fetch
    return @rss if @rss
    begin
      rss = RSS::Parser.parse(rsssource)
    rescue RSS::InvalidRSSError
      begin
        rss = RSS::Parser.parse(rsssource, false)
      rescue  RSS::NotWellFormedError
        rss = nil
      end
    rescue  RSS::NotWellFormedError
      rss = nil
    end
    @rss = rss
    return @rss
  end

  def title
    return @title unless @title == ''

    @rss = rss if @rss == nil
    return 'unknown' if @rss == nil
    begin
      @rss.output_encoding = @output_charset
    rescue RSS::UnknownConvertMethod
      return ''
    end

    @title = @rss.channel.title
    return @title
  end

  def output(num_items=20)
    output = ''
    @rss = rss if @rss == nil
    return output unless @rss
    begin
      @rss.output_encoding = @output_charset
    rescue RSS::UnknownConvertMethod
    end
    @title = @rss.channel.title

    num = 0
    @rss.items.each do |item|
#      date = item.dc_date ? Time.at(item.dc_date).strftime("%y-%m-%d %H:%M") : ''
      date = item.pubDate ? Time.at(item.pubDate).rfc2822 : ''
      line = @output_format.gsub(/%.+?%/) {|p|
        case p
	when '%N%'
          num + 1
        when '%DATE%'
          date
        when '%URI%'
          item.link
        when '%TITLE%'
          item.title
        when '%DESCRIPTION%'
          item.description
        when '%CONTENT_ENCODED%'
          item.content_encoded
        end
      }
      output << line

      num += 1
      break if num_items <= num
    end

    output
  end

end

############################################
# main 
############################################

cgi = CGI.new('output4')
File.umask(0066)

num_items = cgi.has_key?('num') ? cgi['num'].to_i : 20

output = ''
feed = Feed.new(cgi['uri'])
output << feed.output(num_items)

if cgi.has_key?('charset')
  feed.output_charset = cgi['charset']
end

header = {'type'=>"text/plain", 'charset'=>"#{feed.output_charset}"}

puts cgi.header(header)
puts output