#!/usr/bin/env ruby
# WebAgent class
#
# Copyright 2001 TAKAHASHI 'Maki' Masayoshi <maki@open-news.com>
# You can redistribute and/or modify it under the same term as Ruby.
# cf. http://www.ruby-lang.org/en/LICENSE.txt

require 'net/http'
require 'cgi'

#require 'simpleuri'

class CGI
  class Cookie

    def Cookie._load(dumped)
      str = Marshal.load(dumped)
      load_str(str)
    end

    def Cookie.load_str(str)
      str =~ /^([^=]*)=([^;]*)(;(.*))?$/
      name, values_str, rest = $1.strip, $2, $4
      name = CGI::unescape(name)
      values_str ||= ""
      values = values_str.split('&').collect{|v| CGI::unescape(v)}
      cookie_hash = {'name'=>name, 'value'=>values}

      if rest
	rest.split(';').each do |pair|
	  name, value_str = pair.split('=', 2)
	  name = CGI::unescape(name.strip)
	  value_str ||= ""
	  value = if name == 'expires'
		    require 'parsedate'
		    Time.gm(*ParseDate::parsedate(value_str)[0..5])
		  elsif name == 'secure'
		    true
		  else
		    value_str
		  end
	  cookie_hash[name] = value
	end
      end
      Cookie.new(cookie_hash)
    end

    def _dump(limit)
      dumped = self.to_s
      Marshal.dump(dumped)
    end

  end
end


class WebAgent

  ## cookie保存用ファイル
  COOKIE_FILE = '.agent_cookie'

  attr_accessor :code, :message, :header, :body
  attr_accessor :cookies

  def uri=(uri)
    case uri
    when String
      @uri = URI.create(uri)
    else
      @uri = uri
    end
  end

  def initialize()
    @code = nil
    @body = nil
    @header = nil
    @old_uri = []
    @cookies = Hash.new()
    @cookie_file = COOKIE_FILE
  end

  def get_http_adapter()
    if ENV['http_proxy']
      @p_uri = URI.create(ENV['http_proxy'])
      return Net::HTTP::Proxy(@p_uri.host, @p_uri.port)
    else
      return Net::HTTP
    end
  end

  def load_cookie
    if @cookie_file && FileTest.exist?(@cookie_file)
      begin
	open(@cookie_file){|f|
	  @cookies = Marshal.load(f)
	}
      end
    end
  end

  def set_user_header()
    @user_header = Hash.new()
    load_cookie()
    send_cookies = @cookies.collect{|key, val|
      path_length = val.path.length
      #p val.path[0, path_length]
      if val.domain == @uri.host &&
	  val.path == @uri.path[0, path_length]
	"#{key}=#{val.value.join('&').gsub(/\+/,'%2B')}"
      end
    }
    @user_header['cookie'] = send_cookies.join("; ")
    @user_header['user-agent'] = 'HTTP-getter (Ruby)'
  end

  def get()
    net_http = get_http_adapter()
    set_user_header()
    net_http.start(@uri.host, @uri.port) do |http|
      request_path = @uri.path
      request_path += ('?' + @uri.query) if @uri.query
      # p [request_path, @user_header]
      res = http.get2(request_path, @user_header)
      @code    = res.code
      @message = res.message
      @header  = Hash.new()
      res.each{|key, value|
	@header[key] = value
      }
      if @header['set-cookie']
	# p @header['set-cookie']
	@cookies = Hash.new()
	cookie = CGI::Cookie.load_str(@header['set-cookie'])
	@cookies[cookie.name] = cookie
	#p @cookies
      end
      # p @header
      @body = res.body
    end
    save_cookie()
    if %r|<meta http-equiv="refresh".*url=([^"]*)"|i =~ @body
      url2 =  CGI::unescapeHTML($1)
      if %r|^/| =~ url2
	buff = @uri.scheme + '://'
	buff << @uri.userinfo << '@'  if @uri.userinfo
	buff << @uri.host
	buff << ':' << @uri.port  if @uri.port
	buff << url2
	url2 = URI.create(buff)
      end
      # p ['url2', url2]
      reget(url2)
    elsif @code =~ /^3/
      location = @header['location']
      reget(location)
    end
    save_cookie()
    @old_uri.clear
  end

  def save_cookie
    File.open(@cookie_file, 'w'){|f|
      Marshal.dump(@cookies, f)
    }
  end

  def reget(uri)
    new_uri = URI.create(uri)
    if @old_uri.include? new_uri
      raise 'loop occured.'
    end
    @uri = new_uri
    @old_uri << new_uri
    get()
  end

end  ## class WebAgent